In the second piece, I wrote about the middle of the organization.
The people asked to carry out change are often the first to feel the contradiction between what the executives want, and the harsh reality of making it happen.

This final piece is about the system.

System is a word that can sound cold.

I do not mean an abstract machine floating above people’s heads. I mean the working conditions that shape behavior every day: decision rights, incentives, resource allocation, roles, capabilities, priorities, consequences, status, routines, and the informal rules people learn by watching what gets rewarded.

This is where change becomes real.

Or disappears.

Leadership impact at system level

If you read my past reflections, you know I care about leadership behavior. Tone, presence, language, listening, courage, and consistency are things that matter.

But I become a bit cautious when leadership gets reduced to interpersonal performance.

I often remind my clients that while better meeting, better conversations, storytelling or listening sessions are useful tools, they are not the whole job.

Leadership in change is personal impact, at system level.

Senior leaders change the system whenever they alter what the organization can decide, fund, stop, reward, tolerate, escalate, learn, and repeat.

Their words matter because people compare those words with these choices.
This is why employees watch senior leaders with such precision during change.

They may not say much, but believe me, they are collecting evidence.

Who gets promoted?

Which project receives resources?

Which conflict is resolved?

Which old priority is finally killed?

The organization reads these signals as the real change narrative.

A leader can say ‘accountability’ twenty times.
Then, when a decision becomes politically uncomfortable, they either clarify ownership or let ambiguity protect everyone. The second choice speaks louder.

A leader can say ‘collaboration’ with great warmth.
Then, when two functions fight over targets, they either redesign the tension or reward the better political player. Again, the organization learns.

A leader can say ‘customer-first’ until the phrase loses all texture.
Then, when internal processes harm the customer, they either remove the internal obstacle or ask the customer-facing team to be more… resilient [Ugh..]

This is why I prefer to ask leaders a practical question during change:

What will people be able to do after this change that they cannot do today?

If the answer is only ‘understand the strategy better,’ we’re not there yet.

Share

Start with the business problem

A lot of change work begins too late.

What I often see is that change management is called in not to define the solution [new structure, new operating model, new performance expectations and metrics, new leadership behaviors, and so on] but just to try and help people accept it.

This sequence creates avoidable pain.

When the solution arrives before the problem has been made clear, people experience change as an imposition.

People are smart. They may understand parts of it, but they struggle to see why this solution, why now, and why this cost.

A stronger approach starts with the business problem in plain language.

‘We are too slow because decisions need to be constantly escalated.’

‘We are losing customers because functions protect their own metrics.’

‘We cannot execute the strategy because leadership roles overlap and nobody owns the full outcome.’

‘We have promoted technical excellence into leadership roles without giving them the tools to succeed in a completely new role’

‘We keep asking for innovation but we have so many rules chocking any deviation from the norm.’

Plain language does a lot of work, because it strips away ceremony and gives people a chance to actually discuss change.

It also creates discipline for senior leaders.

Once the problem is clear, the organization can ask whether the proposed solution actually addresses it.

If decisions sit too high, will the change move authority downwards? If functions protect their own metrics, will incentives and performance conversations change? If leadership roles overlap, will decision rights become clearer? If innovation is blocked by risk habits, will risk appetite be redefined in practice?

This prevents the common tragedy of solving the stated problem while preserving the actual cause.

Plenty of organisations restructure in the name of speed and keep every approval intact. They launch a new culture while continuing to promote the old leadership profile, or announce empowerment, then punish the first person who takes it seriously.

As someone told me once: Sometimes, organizations look like that guy who was painting the entrance door while the kitchen was on fire [or something like that]

Share

Name the trade-offs early

Every meaningful change creates loss.

Speed often asks local teams to give up some control.

Enterprise accountability can reduce functional autonomy.

Customer focus may make internal comfort less protected.

A leaner structure can narrow career options, while stronger prioritisation will almost certainly disappoint stakeholders who were used to being accommodated.

More transparency may expose weaker leaders, and a clearer operating model can remove the ambiguity some people have learned to use rather well.

These are what I normally call ‘trade-offs’.

Companies often avoid discussing these out loud because they fear making people anxious.

I’ll tell you a secret.

People are already anxious.

Trade-off conversations are useful because they move the change from aspiration to consequence.

Senior leaders should be able to say what will become harder, who will lose discretion, which work will stop, where the organization will absorb pain, and which tensions are intentional.

That last phrase is important: intentional tensions.

Let’s be clear on this: Some tensions mean the change is badly designed.

But others mean the strategy is finally touching reality.

The organization needs to know the difference.

For example, a company moving from local autonomy to enterprise customer management will create tension between country priorities and global account decisions.
That tension can be healthy if decision rights are clear and incentives support the new logic. It becomes destructive when leaders keep praising enterprise behavior while rewarding local protection.

A company moving toward faster innovation will create tension between experimentation and control.
That tension can be productive if risk boundaries are explicit. It becomes lip service when every experiment still requires the same approval logic as a capital investment.

A company asking for leadership accountability will create tension between ownership and consensus.
That can be useful if roles are clear. It becomes cruel when people are held accountable for outcomes they cannot influence.

Trade-offs make change real. Hidden trade-offs make change political.

Leave a comment

Remove the old requests

Companies are much better at adding than subtracting.

They add:

• Priorities

• Behaviors

• Values

• Initiatives

• Governance

• Platforms

• Leadership expectations

• Capability programs

• [And sometimes they add dashboards to explain the other dashboards]

Then they wonder why people seem a bit too stretched.

A serious change subtracts.

• It removes old work.

• It retires old metrics.

• It reduces approvals.

• It stops reports that no longer serve the strategy.

• It kills useless meetings.

• It changes resource allocation.

• It clarifies which decisions no longer require consensus.

• It tells people which previous expectations have expired.

Subtraction is one of the most underused forms of respect in organizations.

People can take on difficult work when the organization creates room for it.
They become cynical when the new world is added to the old world and sold as empowerment.

At some point, the company is basically asking people to become… taller.

Subtraction is also a test of strategic seriousness.
If nothing can be stopped, the strategy has no priority - at best, it has branding.

If you’re keeping the old system just adding a new system on top of it, the old system will win.

And you know why?

Because the old system most of the time has payroll on its side.

Subscribe now

Build feedback loops … with teeth

Feedback is one of those words that often becomes just decorative.

A company collects feedback, then life continues.

Feedback loops only matter when they can change something. I call them ‘feedback loops with teeth’.

They do not need to change everything, sure. Adults can understand this.

What people need is evidence that their input has somehow entered the operating system.

A proper feedback loop answers four questions.

1. What did we hear?

2. What will we do because of it?

3. What will we keep unchanged, and why?

4. What remains unresolved?

That fourth answer is useful because it avoids fake closure.

Many changes contain live tensions, and naming them honestly can build more credibility than pretending everything has been solved.

Feedback with teeth also protects the quality of the change, because it reveals where the design is too abstract, where capability is missing, where old processes are blocking new behavior, where customers may be affected, where middle managers are exposed, and where senior decisions are needed.

By doing this, you’ll finally move feedback out of the ‘meaningless kindness’ category and into the ‘intelligence gathering’ category.

This is where psychologically safer environments matter [I use the phrase carefully because it has been softened by overuse].

People need enough safety to say useful things.
That means the truth can travel without being punished on arrival.

Some companies say they want candor, but far too often what they want is candor with no implications.

Real candor, unfortunately, creates work. I’ll push it one step further: Candor demands work.

Share

Make readiness concrete

Readiness is another word that gets abused.

In weak change work, readiness becomes sentiment.

• Are people positive?

• Are they engaged?

• Do they support the change?

• Do they understand the message?

These questions have value, but they are also incomplete.

A more useful version of readiness asks whether the organization is actually able to implement the change.

• Do people have the required skills?

• Are decision rights clear?

• Are resources available?

• Do incentives support the new direction?

• Are senior leaders resolving contradictions?

• Are middle managers equipped?

• Are old priorities being removed?

• Are customers protected?

• Are risks named?

• Are local teams allowed to adapt implementation without corrupting the intent?

This makes readiness operational, and less flattering.

Many organizations prefer to measure whether people are ‘on board’ because that keeps attention on employee attitude. Operational readiness puts attention back on leadership choices and system design.

And that is good.

Attitude matters, of course, but conditions matter more than most leaders want to admit.

This is where the better change models can help, provided they are used lightly.

They give us useful prompts: check whether people understand the change, want to support it, know how to act differently, have the ability to do so, and see reinforcement from the system around them.

They push us to think about coalitions, the normalization of new practices, and the gap between intention and behavior.

At their best, they also remind us that shared commitment and shared confidence are built through real conditions, not enthusiasm alone.

The models are helpful servants, but they make poor landlords.

Leave a comment

The practical discipline

If I had to describe change management in the least glamorous way possible, I would call it

the discipline of making a strategic change executable by real people inside a real system

This definition will not sell many conference tickets, I’ll admit.

But I like it because it leaves fewer escape routes.

• It means the strategy has to be clear enough to guide choices.

• The operating implications have to be named.

• The people closest to the work have to be involved early.

• Leaders have to stay connected to consequences.

• HR has to work near to reality.

• Middle managers need authority and protection.

• Feedback needs to change decisions.

• The old system has to surrender something.

It also means change cannot be replaced by the change plan.

A plan can coordinate work, sequence activities, clarify responsibilities, and it can be very useful to help people see progress.

But the plan is never the change.

The change is the accumulation of altered decisions, repeated behaviors, adjusted consequences, removed obstacles, learned capabilities, and new patterns that keep their shape under pressure.

This is very important because real change needs to work after we are done talking about it.

If your culture change collapses during budget season, it means it didn’t work.

Similarly, a leadership change that disappears around powerful executives has told people exactly where the boundary is.

Real change survives ordinary pressure.

It does not survive because people were inspired in week one, but because the organization has made the new behavior easier, safer, more useful, more rewarded, and more expected than the old behavior.

Again, unromantic. But gain, hard to fake.

Subscribe now

A small manifesto, unfortunately

Since this is the final piece, I will risk a small manifesto.

I’ll make a slightly joking one, because if you made it this far, a dramatic one would probably be too much.

Change begins after the announcement.

Communication earns attention.
Conditions earn behavior.

A model can guide thinking.
It cannot replace judgment, courage, or consequence.

People do not resist only because they dislike change.
They often resist because the organization has asked them to believe words and ignore evidence.

Middle managers are not delivery pipes.
They are interpreters of reality. Treating them as pipes is a good way to damage both the people and the change.

HR should never be asked to make a message true without access to the business levers that shape truth.

Senior leaders do not need to perform constant visibility.
They need to make choices that prove the strategy has entered the system.

Participation should have boundaries, authority, and consequences.
Otherwise, it becomes a nice, but useless and fleeting, staged performance.

Feedback should change decisions, clarify constraints, or expose unresolved tensions. Appreciation alone is too cheap.

The old work must be removed.
The old incentives must be questioned.
The old permissions must be rewritten.
The old habits must lose some of their protection.

If everything remains possible, nothing has been prioritized.

If nobody powerful is inconvenienced, the change is probably still polite.

If the new behavior becomes rational, supported, and expected under pressure, then the organization may finally be changing.

That is the work.

Less shiny than a launch video.

But more useful.

Read part 1 and part 2 of The Illusion of Change series!

This second piece starts where the announcement usually lands.

Somewhere in the middle.

The middle of the organization is where change stops being conceptual and starts becoming practical, emotional, political, and occasionally absurd.

Strategy turns into workload. A new operating model becomes a difficult conversation with someone whose role has just become unclear. ‘Faster decisions’ runs into a senior leader who still wants to approve everything, and ‘empowerment’ meets a policy last reviewed when dinosaurs were still roaming our planet [very bureaucratic dinosaurs, I know].

The middle is also where companies often place their change agents.

I like change agents.

I work with them often. In the best cases, they are translators of the future. They understand the formal change and the informal system. They can tell you where the plan will jam, who will resist for intelligent reasons, which phrase will trigger suspicion, and which leadership behavior has already created a credibility problem.

Useful people.

Then many companies do something rather unfair.

They nominate these people, brief them, energize them, ask them to engage the organization, and leave them standing between promise and reality with very little authority.

This is how a good idea becomes a human shield.

Share

The human shock absorber

I once worked on a change where we designed a series of activities to involve the people most affected by a new direction.

The change agents were mostly middle managers and functional representatives from across the business. Smart, busy, nominated by the organization.

The intention seemed fairly sensible:

• help them understand the change,

• equip them to engage colleagues,

• create feedback loops, and

• surface real barriers before those barriers became expensive.

The senior leaders approved the approach and signed off the communications.
After that, their involvement began to thin.

Nothing dramatic happened.
No villain walked out of the room wearing a black cape [unfortunately].

The signs that things were going sideways were much less flamboyant: decisions that should have changed stayed exactly where they were, follow-up became vague, and trade-offs kept being pushed into a later conversation.
[I feel a vague sense of dismay when I hear ‘Let’s take this offline’]

Feedback from the change agents was heard, appreciated, and then placed gently into the organizational fog.

Share

The problem is, the change agents noticed it.

At first, they responded by working harder.
They explained the change with more patience, reassured teams, collected questions, and tried to make sense of the ambiguity for others.
In practice, they were absorbing frustration that belonged several levels above them.

Then the mismatch became too large.

The organization had asked them to represent a change that the senior team was only partially enacting.

That is a brutal position. It turns middle managers into human shock absorbers.

They carry the emotional load of decisions they did not make, the ambiguity of trade-offs they cannot resolve, and the credibility risk of promises they cannot keep.

Eventually, exhaustion appears. Followed by irony and then, inevitably, by silence.

Silence is often misread as acceptance. In change work, silence can also mean people have updated their risk calculation, and found out the risk is no longer worth it.

Subscribe now

The myth of the messenger

There is a persistent fantasy in organizations that a message becomes stronger when repeated by enough people.

To be fair, sometimes it does. A message does become stronger when it is repeated consistently and translated well at local level.

The fantasy begins when repetition is expected to compensate for weak evidence.

A middle manager can explain why the strategy is changing, help the team understand the implications, surface concerns, and translate corporate language into something closer to work. A good one can also hold tension with maturity, at least for a while.

But there are limits.

They cannot make unresolved decisions disappear, explain away incentives that contradict the message, or create authority they have not been given.

They cannot protect employees from consequences that senior leaders refuse to name.

And for sure, they cannot keep saying ‘we are serious this time’ while everyone sees the same behaviors being rewarded.

After a while, the messenger becomes contaminated by the message.

This is one reason employees become skeptical of local leaders during change.

The local leader may have done nothing wrong. Very often, they are doing their best with partial information, limited authority, and a script carefully designed to avoid difficult conversations.

But they are the person in the room, so the frustration attaches to them.

Meanwhile, the senior team remains conceptually supportive and physically elsewhere.

[A very efficient arrangement… for the senior team]

Leave a comment

HR in the ventriloquist position

HR often gets caught in the same trap.

In many organizations, HR is asked to carry the human side of change without being given sufficient influence over the business conditions that create human behavior.

HR can improve the narrative, equip managers, facilitate engagement, design listening sessions, support role transitions, build capability plans, and create feedback channels.

Good HR can do all of that with skill and courage.

But again - just like with middle managers, there is a limit.

HR cannot, alone, change what the senior team continues to reward.
It also cannot compensate forever for unresolved strategic trade-offs or keep asking middle managers to carry a message that leaders are unwilling to back through resource allocation, role clarity, incentives, and real consequences.

When HR tries, the function becomes a kind of organizational ventriloquist.

The voice sounds senior, but the body in the room belongs to HR.

And everyone can see the trick, which inevitably looks cheap.

This is painful because HR is then blamed from all sides.

Leaders feel HR is too soft, too slow, too concerned with reactions.
Employees feel HR is selling a message it cannot make true.
Middle managers feel HR has given them tools without enough cover.

Sometimes everyone is partly right.

The issue, in my opinion, is structural.
HR is often asked to manage response without being close enough to the levers that shape reality.

That creates beautifully facilitated powerlessness.
[A niche but surprisingly common product]

Subscribe now

For change to work, HR needs access to the decisions that define the change:

• roles,

• governance,

• incentives,

• performance expectations,

• talent consequences,

• capability investment,

• sequencing, and

• the removal of old work.

Otherwise, HR becomes the department of tone.

Tone matters, sure. But reality tends to win.

Share

Why people resist the wrong thing

Companies often talk about resistance as if it lives somewhere inside people.

There are several archetypes in this tragic novel:

• The resistant manager.

• The skeptical employee.

• The difficult function.

• The negative team.

Sometimes this is accurate.

Some people resist because they are protecting status, comfort, identity, or influence. Others resist because the change has been badly explained, because uncertainty feels costly, or because the old world worked rather well for them.

Yet a lot of what gets labelled resistance is actually people responding rationally to contradiction.

They hear that the company wants speed, while decision rights still move slowly through the same old channels.

Empowerment is presented as a priority, but every meaningful choice climbs back upwards.

Collaboration is praised, while leaders continue fighting for functional territory.

Talent is supposed to be assessed differently, yet the same old profiles keep getting promoted.

In that environment, skepticism may be described as a mood problem, but I think that most often is more like pattern recognition.

This is where behavioral science is useful because it brings the conversation back to conditions.

People are more likely to behave differently when they have the capability to do so, the opportunity to do so, and enough motivation to sustain the effort.

Remove one of those, and the change becomes fragile. Remove two, and the change becomes just an illusion [with a feedback survey in the best cases]

Subscribe now

Capability is often underbuilt.

Companies assume that because people understand the new direction, they can operate inside it. This is rarely true. A new strategy may require different commercial judgement, cross-functional negotiation, leadership maturity, data habits, customer conversations, risk decisions.

These are skills, not vibes.

Opportunity is often blocked.

People may want to behave differently, but the structure, workload, authority, governance, or technology prevents them. This is where organizations become strangely moralistic.

They ask for ownership from people who do not control the conditions of ownership.

Motivation is often treated as a communication problem.

Leaders believe people will care more if the case for change is compelling enough. A strong case helps, of course. Sustained motivation usually comes from seeing that the organization is serious enough to change consequences.

This brings us back to the middle.

Middle managers see these conditions first. They know when capability is missing, when opportunity is blocked, and when motivation is being drained by contradiction.

A smart change approach treats them as diagnostic partners.

A weak approach treats them as message distributors.

The difference is expensive.

Leave a comment

Participation without authority

Participation is another area where organizations can accidentally create cynicism.

I believe in involving people in change. Strongly.

People closest to the work often understand the practical system better than the people designing the change.
They can identify friction, hidden dependencies, customer implications, capability gaps, emotional risks, and political landmines. Ignoring them is wasteful.

Still, participation has to be honest.

When people are invited to shape something that has already been decided, the organization should say so.

• When input is welcome only on implementation details, say that.

• When the business direction is fixed but the sequencing is open, say that.

• When feedback will be considered but cannot override a financial constraint, say that too.

Nobody needs a fake democracy, but everybody will be delighted to see clean boundaries.

The worst version of fake democracy is performative inclusion.

Workshops are held, sticky notes appear, and concerns are captured with great care. For a moment, people feel involved. Then nothing visible changes, or the final decision looks remarkably similar to the original plan. The organization thanks everyone for their input and moves ahead.

This teaches a lesson which is possibly the most absurdly counterproductive lesson anyone, and least of all organizational leaders, would need to teach anyone else.

People learn that participation is ceremonial, so next time they become more careful.

They say less, wait longer, and give the organization less to work with.
The organization then notices lower engagement and [with impressive optimism] designs another listening exercise.

At this point, satire becomes difficult because reality is already doing the work.

Good participation has authority attached to it.
The authority may be limited, but it has to be real.

A group may not be able to change the strategic direction, but it can still shape the sequence, identify old processes that need to stop, test customer implications, redesign handoffs, propose clearer roles, or spot capability gaps before someone builds the wrong training.

It can also escalate contradictions that leaders must resolve within a defined time, which is often where participation becomes genuinely useful.

Share

The senior leadership absence problem

Senior leaders do not need to be everywhere in a change.

In fact, when they try to be everywhere, they can create a different problem: Everyone starts performing commitment upward.

The senior leadership role is more strategic, and more demanding, than constant visibility.

Leaders have to stay connected to the change as a living operating issue. That means checking whether the organisation is actually making the new behaviour possible, removing obstacles surfaced by the business, resolving conflicts between old priorities and new expectations, and making choices that cost them something.

That cost is essential.

Let me give you a few examples to clarify what I mean:

When leaders ask for enterprise behavior, they have to reward it even when a function loses something.

Faster decisions require some loss of control.

Accountability requires leaders to stop rescuing others from the consequences of ambiguity they created themselves.

Customer focus requires accepting that internal comfort will take a hit.

Change becomes credible when senior people absorb some of the cost of their own strategy.

This is where visible detachment does so much damage. The organization does not need senior leaders to perform constant enthusiasm, but it needs evidence that the strategy is shaping choices at the leadership level, too – not just where common mortals are.

When that evidence disappears, middle managers are left hanging in there, and this is when change agents become tired.

Their fatigue is rarely caused by too much activity.

It comes from defending a future that the present [and their leaders] keeps undermining.

Share

Where the third piece begins

The middle of the organization should never be treated as a loudspeaker, because it is the place where the change is tested against reality.

Middle managers, HR partners, and change agents can translate, diagnose, support, and challenge.

They can do remarkable work when they have context, authority, protection, and access to decisions.

When they are used as carriers of leadership intent without enough leadership action behind them, the change starts losing credibility from the inside.

The third and final piece of this mini-series will move from people to system.

Because after the announcement, after the middle has absorbed the first wave of confusion, the central question becomes very practical: what must the organization change in its own operating conditions for the strategy to become real?

This is inconvenient, because a large part of my work is exactly that.

Strategy changes. Structures move. Roles disappear, appear, merge, split, or acquire names that make everyone slightly sad.

Senior teams decide that the old way of working has reached the end of its useful life. HR gets pulled in. Someone builds a plan. Someone else builds a nicer plan. Then the company announces the plan.

And very often, this is where the serious misunderstanding begins.

The announcement feels like things are moving. It comes with a date, a cool title, a leader message. Maybe a launch video.

[A dangerous format, the launch video, if you ask me]

People receive the message. They understand, more or less, that something is happening, and they may even agree with the logic.

Markets have changed. Margins are under pressure. The operating model is too slow. The leadership culture rewards escalation and punishes ownership. The strategy requires different behavior.

So far, fine.

Then reality kicks in, normally when the next day people goes back to work.

The approvals still sit in the same places. Decisions still climb back to the same senior people. Incentives still reward local protection, resources still follow old priorities, and talent conversations still happen in the familiar coded language.

The people asked to act differently soon discover that the system has remained stubbornly loyal to yesterday.

This is where many companies lose the change.

They rarely lose it because the communication was badly written. Some communication is dreadful, of course. I have read change messages so smooth they seemed vacuum-packed.

Even a good message has a short shelf life when daily experience contradicts it.

A company says it wants to move faster, while seven approvals still stand between a decision and action.

It talks about accountability, while decision rights remain as foggy as before.

It promotes a more collaborative culture, while leaders who hoard information continue to win.

People notice, and quickly.

Subscribe now

The comfort of communication

I understand why companies reduce change to communication.

Communication is visible.

It can be planned, polished, translated, distributed, tracked, and reported.
It gives leaders the feeling that something has been done. It gives HR a task. It gives consultants a workstream.
Everyone can point at the artefacts: A deck, a timeline, or even a page on the intranet that nobody will ever visit, apart from those who created it.

The seduction is obvious.

Behavior is messier, because it asks irritating questions.

• Can people do the new thing?

• Are they allowed to do it?

• Will the system punish them for doing it?

• Will their leaders tolerate the consequences?

• Will resources move?

• Will priorities be killed, rather than merely joined by younger, shinier priorities?

These are less elegant questions, but they are also the real work.

Most serious change research, whether it comes from organizational psychology, implementation science, behavioral science, or old-fashioned OD practice, keeps circling back to the same uncomfortable truth:

People change behavior when the conditions around them change with them.

Understanding, commitment, leadership, participation, skills, incentives, reinforcement, social proof, psychological safety, and a decent narrative all help. They just help in different ways, and none of them can carry the full weight of change alone.

None of these work well when used alone as decoration.

A message can explain the destination, but it cannot carry the weight of a system that keeps asking people to behave as before.

This is something that companies often underestimate.

They ask people to ‘embrace the change,’ while leaving intact the very conditions that made the change necessary.

It is a strange request. Almost athletic.

Please be faster inside a slow approval system.

Please be accountable inside an ambiguous decision system.

Please be enterprise-minded inside a reward system that still pays you for protecting your function.

Please be innovative inside a culture that treats every mistake as a criminal act.

People are capable of remarkable adaptation, but unfortunately [or luckily], they are also quite good at detecting nonsense.

Leave a comment

The old system has a longer memory

Every organization has a memory.

Part of that memory sits in people’s heads.

A much, much larger part sits in routines, incentives, power patterns, reporting lines, budgeting habits, promotion logic, and the small daily consequences that teach people how things really work.

This is why change announcements often create such a strange emotional mix.

People can feel hopeful and skeptical at the same time.
On one hand, they may genuinely want the strategy to work, while on the other hand they are probably painfully aware of which bbehaviors will never survive once they are enacted in real life.

The leader says:

We need more ownership.

The experienced manager hears, ‘Let us see whether you still escalate every uncomfortable decision back to yourself.’

The CEO says:

We need to become more customer-led.

The commercial team hears, ‘Let us see whether finance, operations, legal, and product will stop killing customer decisions through a thousand reasonable objections.’

HR says:

We want a more open culture.

Employees hear, ‘Let us see what happens to the first person who says the thing everyone knows.’

This is where change becomes behavioral, political, and operational. It leaves the safe world of messaging and enters the actual company.

Good change work respects that complexity. At least in the beginning, skepticism should be treated as data rather than attitude. People resist for many reasons: self-interest, emotion, politics, fear, fatigue, or simply because they can see something the plan has missed.

[That last kind of resistance is precious, by the way - because it tells you where the design is fragile]

Many organizations waste that data because they are too busy managing sentiment.

Leaders sometimes ask for honest feedback and then visibly dislike the honesty.
HR teams collect concerns, soften the edges, group them into acceptable themes, and send them upwards with most of the teeth removed.
Change agents explain issues with real precision, only to receive warm thanks and no meaningful action.

[Warm appreciation is a very elegant way to do nothing]

Subscribe now

The model is innocent

At some point, someone usually asks which change model we should use.

This is a fair question. Models can help. A good model gives people a common language. It reminds the organization that change has stages, stakeholders, emotions, barriers, reinforcements, and a very annoying tendency to go off-script.

I am not against models. I am against hiding inside them.

There are many useful bodies of work behind serious change practice.

Lewin’s old idea of changing the forces around behavior still has value. Kotter’s work on urgency, coalition, and reinforcement remains useful when treated with judgment. ADKAR can help leaders remember that awareness and desire do very little without knowledge, ability, and reinforcement. Behavioral science reminds us that motivation alone is weak when capability and opportunity are missing. Implementation science keeps asking how new practices become normal in real settings, which is exactly the question many change plans avoid.

All useful. All incomplete when treated as recipes.

A model can help a senior team think more clearly, but it will not decide which legacy priority must be killed.

It will not make a powerful leader give up control, protect a middle manager who raises an inconvenient truth, move resources from the old strategy to the new one, or make the reward system less hypocritical.

That work belongs to the organization.

This is one reason I become skeptical when change management is sold as either too small or too grand.
On one side, there is the minimalist version: send the message, train the managers, update the FAQ, job done.
On the other side, there is the cathedral version: complex architecture, branded methodology, colorful maturity levels, a governance structure large enough to require its own governance structure.

Both can miss the point.

The useful work is usually more grounded: a clear business problem, honest trade-offs, real involvement, and authority that matches responsibility.
It means feedback has somewhere to go. Senior leaders stay close to the operating implications. HR works near enough to power to help make the message true. Middle managers are equipped as interpreters, instead of being used as shock absorbers.

Less glamorous. Much harder to fake.

Leave a comment

Strategy has to disturb something

A strategic change that disturbs nothing is probably just a campaign.

Real strategy disturbs the organization.
Money moves. Decision rights move. Some capabilities become more valuable, while others lose status. Certain leaders start to look future-fit, and others suddenly look very attached to the past. Behaviors that used to pay off become expensive. Work that once felt untouchable has to stop.

That last part is where many companies become shy.

They begin doing something silly: Adding the new strategy on top of the old operating logic. New priorities arrive, old priorities remain, and people are asked to reconcile the conflict through personal resilience.

This is how organizations create confusion while calling it ambition.

A restructuring, for example, may redraw the boxes. The real question is whether power, accountability, workflow, and talent logic also move. Otherwise, the company has created a new structure for old behavior.

A leadership culture change may define new expectations. The question is whether those expectations survive pressure. Culture is easy when everything is running smoothly. It becomes visible when a big customer complains, a senior person is challenged, a quarterly number is at risk, or a talented executive behaves badly and still delivers revenue.

A strategy change may announce a new direction. The question is whether the organization changes the conditions that made the old direction rational.

That word matters: rational.

People often behave in old ways because the old system still makes those behaviors rational.

If escalation protects you, you escalate.
If hoarding information gives you power, you hoard.
If local targets define success, you defend your local target.
If challenging a senior leader creates risk without reward, you learn to phrase disagreement as a question and then eventually stop asking.

This is less about attitude than companies like to admit. It is about the logic of the system.

Change work becomes serious when leaders ask:

What are we currently making rational that we now claim we want to change?

That question tends to reduce the number of posters.

Subscribe now

The first adult conversation

The first adult conversation in change work is usually about cost.

I do not mean budget, although budget matters. I mean organizational cost.

• What will become harder?

• Who will lose discretion?

• Which habits will be interrupted?

• Which group will carry more complexity?

• Which leader will lose a familiar source of power?

• Which metric will create tension with the new direction?

• Which promise will need to be walked back?

This is where many change plans become evasive.

They describe the benefits in detail, while the costs remain conveniently misty. The future is sold to people who are still dealing with the present, and belief is requested before the organization has produced enough evidence to deserve it.

People can handle difficult news better than vague optimism.

In fact, vague optimism is often more corrosive because it forces employees to do the emotional accounting themselves. And this accounting rarely ends up well for the leaders pushing vague optimism.

A more serious approach says: here is the business problem, here is the direction, here are the trade-offs, here is what we know, here is what we do not yet know, here is where we need input, here is what will change as a result of that input, and here is what has already been decided.

There is nothing magical here. It is simply adult.

[I know, adult can be scary]

Leave a comment

That’s it for part 1. Stay tuned for part 2

The announcement is the cheap part because it can be completed before the organization has truly changed anything.

The expensive part begins when people try to act on the message and discover whether the system will let them.

That is where the middle of the organization becomes… central [pun not intended, I promise].

In the next piece, I want to look at the people who are usually asked to carry the change after the announcement: middle managers, nominated change agents, HR partners, functional representatives, and all the practical adults who end up translating executive intent into organizational reality.

That’s an unfair job. Too often, these people end up being called blockers, while they are just trying to absorb contradictions.

For years, I’ve been hearing hundreds of variations of the ‘we should invest in people’ line.

[A line that most of the times means ‘let’s buy a course and call it a day’]

But these days, I can sense real excitement about how AI can finally make development faster.

If you’ve ever sat through a talent review where everyone agrees that

we need more development

and then nobody has time to do anything meaningful about it, you’ll feel the appeal in your bones.

But now, AI shows up and suddenly you can produce:

• skill roadmaps

• development plans

• learning pathways

• talent summaries

…at scale.

On one hand, it’s an undeniable breakthrough. But on the other hand, I keep having the same, slightly uncomfortable, thought.

If AI makes development plans too quickly and too well-written, it becomes very easy to confuse output with progress.

[Which is a polite way of saying: we may end up industrializing nonsense]

This is the third and last piece in my mini-series on AI for HR: Recruitment, Performance, and now Learning & Development.

This stage of the Talent/HR cycle is very important, of course, but also in my experience the most overlooked.

Talent Development is not just training.
Training is a tool.

The real thing about Talent Development is that it has a direct effect on a number of crucial aspects of organizational life - and business.

It impacts performance, retention, engagement, etc etc.

But at the same time, it’s where we decide who gets better - and who gets opportunities.

The real temptation: speed and scale, instead of relevance

Here’s what’s going on with most organizations I am seeing:

First, they discover that AI can generate training plans in minutes.

Then, the conversation immediately becomes:

How do we scale this across the whole organization?

Fair question, but also wrong first question.

Because what AI should really buy us is not just speed.

It should buy us space.
More time to target initiatives and connect learning to real work.
But also more space to talk to managers and ask:

What problem are we solving, exactly?

Instead, I’m watching teams use AI to go faster… in the same direction they were already going.

And if the direction is wrong, faster is just more expensive.

Share

A confession: L&D has been ‘box-ticking’ long before AI

Before we blame AI for everything, we need to be honest about what L&D often looks like.

A lot of companies treat training in one of two ways:

1. Training as the solution to everything
   Team collaboration issues? Training.
   Low accountability? Training.
   Weak strategy execution? Training.
   [At this point, training is basically corporate vitamin. Fancy but most likely useless]

2. Training as a checkbox
   We did the program.
   We had attendance.
   People rated it 4.6/5.
   Everyone can go back to work now.
   [And forget everything they had to endure through the training, the moment they open their first email]

Let me be very clear: AI doesn’t create these habits. These habits have been around since forever - but AI makes them easier to scale.

And that, is the real problem.

Subscribe now

The ‘catalogue shopping’ problem [now with turbo]

There’s another habit that I am painfully familiar with.

Organizations go shopping for training like they’re picking snacks at a convenience store.

Emotional Intelligence had its era.
Then the Change Leadership bonanza.
Then Agile everything.
Now a lot of Resilience and Psychological Safety poorly packaged to hide the fact that businesses know their people are unhappy, but have no idea on how to fix that.

Now, these are all well-marketed concepts, but in many cases, they’re barely thought through.

Don’t get me wrong. They are useful, they can truly help, but they’re just imported as cool ideas, instead of being designed as interventions.

AI makes this worse in a very predictable way:

Anyone can hop on [insert your favorite LLM] and prompt it to generate a 12-month learning roadmap on Agility.
You’ll surely get something that looks credible and that includes all the needed buzzwords. You can sell it to your leaders.

And because it’s clean, structured, and supersonic fast, I can guarantee you - 9 times out of 10 no one will ask:

Agility for what? In which roles? To solve which business constraint? With what trade-offs?

That pause is where L&D becomes strategic.

AI removes the pause - but only if if you let it.

Leave a comment

What ‘too good, too fast’ looks like in development

In a training session recently, someone said something along the lines of:

Now we can finally give everyone an IDP

Sure.

But there are two questions hiding behind that statement:

• Will the IDP change what happens at work?

• Will the manager sponsor any of it?

Because if the answer is no, then we’re producing documents, not development.

And AI is extremely good at producing documents.

AI can do wonders if you need to turn vague, weak inputs into plans that sound thoughtful, or if you need to produce language that sounds fair.

The risk is that you are sacrificing capability on the altar of great-looking presentations.

[Also, please allow me a slightly contentious note here: If your organization has been unable to do development planning for years, and now it suddenly can because a tool writes faster… the bottleneck was never writing.]

So what should AI be doing in L&D?

In the course design that sits behind this mini-series, the development module is framed around this foundational concept:

We should use AI to personalize development at scale while keeping judgment and governance with HR.

Translated into normal human language: Let the tool help you draft and structure, but HR has to own the logic of the bet.

Because every development plan is a bet on:

• where the business is going

• what the role really requires

• what the person can realistically build

• what opportunities will exist

• what the manager will actually support

AI cannot own any of the above, unfortunately.

Share

Slow down, and practice

These are the habits I like because they slow you down when it actually matters.
Word of warning: they’re mildly annoying. That’s usually a good sign.

Habit 1: Start with a role reality check

Before generating anything, HR and the manager should sit down and write two short paragraphs:

• What this role needs to deliver in the next 6-12 months

• What capability will make the biggest difference

• Optional but recommended: If there’s a competency model in your business - what competencies should be considered as predictor of success [in connection to the first point - goals for the next 6 - 12 months]

Then, and only then, use AI.
[Pro tip: Structure a solid prompt, give it the source content, establish its constraints - be deliberate in your use of AI - don’t let it keep its thinking in a black box]

This prevents the classic outcome: a plan that matches the job title and misses the job.

Habit 2: Force a choice between three developmentpaths

Ask AI to generate 3 options. Something like:

• conservative path - incremental growth

• stretch path - bigger growth, higher risk

• role-pivot path - building capability for a plausible next role

Then the manager and HRBP must pick one and write why.
[Pro tip: No ‘we will do a bit of everything’. That’s how you get a plan that does nothing]

No “we will do a bit of everything.” That’s how you get a plan that does nothing.

Habit 3: Separate skills from opportunities.

If the plan is all courses, it’s probably not a plan.

Have AI propose:

• 2–3 skills - Specific skills, not just buzzwords
  [say no to single-word skills, say yes to behavioral descriptors]

• 2 opportunities - E.g. projects, rotations, leading something real, solving a real problem

• 1 relationship - E.g. mentor, shadowing, peer coaching

Then, sanity check: Are these skills, opportunities, and relationships actually realistic and available, or are we writing fiction?

Habit 4: make evidence non-negotiable

It takes five minutes to ruin a monthly checkpoint.

• one example of new behavior

• one example of improved outcome

• one adjustment

AI can summarize evidence, and it’s absolutely great at it, but it cannot see it.

A manager’s job is to find development oppotunties and propose plans - but after that, it’s their responsibility to observe, monitor and gather evidence.

[I would push it as far as to say: gathering evidence for their people’s development should be included in the manager’s performance targets. Harsh, but who else should be accountable for people’s development journey, if not the person who sees their work every day?]

Wrapping up the series

Recruitment. Performance. Development.

Three parts of the HR cycle where AI is genuinely useful, and where HR can lose capability if the tool becomes a substitute for thinking.

In recruitment, AI can help you move faster. It can also help you mis-hire faster.
In performance, it can make reviews sound better than the evidence behind them.
In development, it can scale plans that feel relevant… until people go back to their desk.

So here’s the question I’m leaving you with:

Are you using AI to go faster, or are you using it to do better work?

Because if AI gives you time back and you spend it producing more documents, nothing changes.

If AI gives you time back and you spend it on:

• better diagnosis

• sharper targeting

• more honest conversations with managers

• more deliberate opportunity decisions

…then you’ve got something.

AI is not the strategy. HR capability is the strategy. Still.

This concludes my 3-part series on AI and HR - based on my current experience delivering HR development programs where we use AI to make HR more strategic - instead of adopting AI tools that make HR redundant.

If you’re interested, you can check more about my AI for HR training here.

And here are the previous articles in the series:

As businesses reopened and routines returned in what someone insists on calling the ‘new normal’, many organizations seemed to rely on muscle memory.

To put it rather bluntly, people still needed to run teams, and decisions still needed to be made, possibly fast.

Urgency won. Deliberate assessment and planning lost.

Over the past few months, though, the shift has been visible.

Leaders are taking action again, and people development is back on the agenda.

Which made me ask a simple question: why now?

Why the comeback is happening now

Every company has its own story, but most of what I’m hearing can be grouped into three drivers.

They overlap, and they feed each other.

1) Skills gaps are suddenly harder to ignore

AI has made capability gaps [current and future] painfully visible.

Even in stable teams, leaders are realizing that new competencies are required to stay relevant and to keep good people.

The interesting [and somewhat uncomfortable] aspect of this is the timeline: The problem here is a ‘right now’ problem, not a ‘two-years-from-now’ one.

Let me make a recent example.

In a roundtable with HR leaders in Shanghai last week, someone described a shift in their software engineering teams.

Engineers who used to spend most of their day coding are now spending a growing share of their time orchestrating and supervising AI agents.

Their job profiles, on paper, haven’t caught up. But their jobs, in reality, already have.

So the question becomes immediate: what does ‘good’ look like for a role today - and what needs to be learned for someone to succeed in it?

Share

2) Hiring stopped being the easy fix

When I first moved to China [and I still had hair on my head], many companies could rely on a simple market dynamic: offer a small bump, and you could usually pull talent from almost anywhere else.

For years, hiring was the default solution to talent problems.

Today, it’s no longer that clean.

This might sound counterintuitive, as unemployment is higher in many segments.

But the real change I see is psychological: in uncertain times, staying put feels safer, and job-hopping slows down. And when fewer people move, even strong companies struggle to ‘fish’ for the right profiles.

At the same time, a younger workforce is showing a clearer preference for non-monetary value: flexibility, time, development, meaning, and social impact.

Compensation still matters, but it isn’t the only lever that works.

In practice, companies are being pushed to take retention seriously, as a system.

And that pulls Talent Management back into the spotlight, because retention requires decisions that must feel both viable and fair.

3) Leaders are tired of defensive mode

After years of reacting, many leaders are recognizing the cost of operating permanently in recovery mode.

They want teams that can adapt, learn, and grow while conditions keep shifting.

That requires mechanisms to identify development needs early and close gaps as quickly as they pop up.

Subscribe now

What I’m seeing on the ground

Across many of the companies I work with [or simply talk with], three themes keep coming up in the way they talk about organizational dynamics: Politics, Friction, and Learning.

Let’s unpack them.

Politics: the ‘visibility’ risk

Companies are revisiting how they evaluate people.

After years of relying heavily on manager observation [often tied to annual performance cycles], a problem has become harder and harder to ignore: people end up being assessed on visibility, not contribution.

That’s the political risk many clients describe.

When performance management is slow, vague, or overly subjective, employees lose trust and start optimizing for optics. And leaders end up making talent decisions using the wrong metrics.

This is why I’m seeing renewed interest in more solid criteria, more frequent check-ins, and data inputs of higher quality.

Friction: collaboration is getting more complicated

With globalization looking less predictable than it did a few years ago, many companies are learning a hard lesson: you can’t simply ‘localize everything’ and assume collaboration will take care of itself.

Whether a company is global, regional, or mostly domestic, friction always shows up in one or more of these three ways:

• slower responses to market shifts

• missed opportunities that competitors catch first

• repeated rework because coordination fails

So leaders are getting more serious about identifying where collaboration breaks: across functions, across locations, across headquarters vs. regions, and sometimes across generations.

Learning: adaptability as a measurable business need

This is the most direct link back to Talent Management.

When roles and priorities change frequently, organizations need to know who can learn quickly, and what support makes learning realistic.

Yes, AI skills are part of it. But I think there’s more to it:

First of all, roles are becoming more fluid, and so responsibilities keep shifting.

And on top of this, the definition [and expectations] of ‘great’ performance is very hard to pin down. Especially if you’re doing annual performance appraisals.

What matters more now is learning capacity, and the organization’s ability to support it without creating panic, fear, or performative upskilling.

Share

Two overdue [and very welcome] returns to the agenda

Two topics are clearly making a comeback in China, both of which were treated as ‘nice to have’ for years.

And it’s genuinely refreshing to see them back, because they tell us something healthier: companies thinking about how to build, not just how to get through the quarter.

Talent Development: back, but better [and more sustainable]

Talent Development is old news, and suddenly urgent again.

But with a twist.

I’m no longer hearing interest in one-off workshops. Instead, more organizations are building longer-term systems designed for retention and application.

For example, some of my clients are asking my team to develop training programs combined with ongoing online group coaching.

By doing this, learners get structured content, then a recurring space to review application with the same coach. They get to discuss and share what they tried, what worked, what didn’t, and what to adjust as new behaviors start to form.

We’re expanding traditional training with ongoing advice, peer support, and knowledge sharing based on practice, instead of theory.

I absolutely love this.

A smaller number of companies are going one step further, and I find it super smart.

They’re creating learners focus groups where people work with a coach not only to reflect, but to co-create ways to embed the learning in the business.

It becomes a sort of lightweight change effort: trained employees turning into practical agents of culture and process improvement.

I’ve seen groups propose ways to cascade learning through their teams. Others have built very concrete process improvement ideas based on what they tested after the sessions.

I am loving this approach because it spells out something crucial: People can do far more than ‘absorb and apply’.

With the right tools and permission, they can actively shape how the organization works.

Share

Talent Diagnostics: data, with a developmental purpose

Diagnostics are gaining momentum, partially because AI has made it easier to work with larger datasets and faster feedback loops.

But what I find very encouraging is the direction: many leaders aren’t asking for data to just police people [or create plans to find out who they can lay-off]. They’re asking for data that can support development, while also giving the organization a clearer view of capability gaps.

One example I’ve been working on with a company founded here in Shanghai: an AI interviewer initially used for 360-degree feedback. The cool thing is, it now produces two complementary outputs: on one hand, individual leadership development reports that people can use as a real starting point for improvement. On the other hand, it generates live, aggregated dashboards that allow HR and leaders to spot competency heat maps, identify priorities, and plan targeted actions across teams, departments, and locations.

Done well, this kind of diagnostic approach turns ‘people data’ into something practical: clearer expectations, smarter development investments, and fewer talent decisions based on guesswork.

A blessing, if you ask me.

Subscribe now

I am very curios to see where things will go from here

The practical takeaway is to treat development less like a quarterly initiative and more like basic operating hygiene: frequent feedback, faster calibration, and learning support that survives the [traditionally disastrous] impact with reality.

Hiring will still matter, but it won’t rescue weak systems.

When the market is cautious and people are looking [and working] for safety, attraction gets harder and retention gets unforgiving. That’s why diagnostics and development are returning together: leaders want fewer decisions based on visibility, vibes, or whoever speaks the loudest.

The organizations that will win are the ones now building a clear definition of ‘good’, measuring it more often than once a year, and giving people a credible path to grow without turning every learning moment into a stress test.

And I have to say, unlike Han in the trash compactor, I have a good feeling about this.

Today, I’m going to zoom in on another very important aspect of the employee lifecycle: Performance management.

Hiring happens in bursts. It’s episodic by nature.

On the other hand, performance happens constantly.
And performance management should be an ongoing activity. Although as we know, it ends up being done quarterly in the best [and yet still insufficient] cases. Worst case: yearly performance appraisal. Stuff from when we had just invented the wheel that somehow, we still deal with.

Now, with AI, we can finally outsource this tedious, difficult and confusing managerial activity to AI.

Fantastic. Or maybe not.

Faster drafts and cleaner language for performance management are tempting. let alone when you can ‘delegate’ Ai to do it for you.

The problem is that AI can improve the paperwork while weakening the judgment behind it.

This article follows the second module of my AI-for-HR training: better performance conversations.

The promise of this module 2 is to learn how to use AI to structure reviews, synthesize inputs, and reduce the time managers spend writing.

But at the same time, it helps HR learn how to avoid a deeper, more fundamental trap: letting the tool do the thinking because everyone is tired.

Subscribe now

The process already struggles with trust

In a lot of companies, performance management is, at best, tolerated.

It’s mostly a matter of admin with some consequences, filled with politics and lots of paperwork so everyone feels a little safer.

Now, AI drops in a system that already has three common weaknesses:

• Thin evidence. Reviews get written from memory and mood because examples aren’t collected.

• Weak manager writing skills. Many managers can’t write clear, specific feedback.

• Vague language habits. People learn to sound professional while saying very little.

AI can help with all three.

But at the same time, it will also unerringly scale the weaknesses if the inputs stay weak.

Subscribe now

AI is trying to be helpful, really

Since performance management is a tedious job, executing it is low on the priority list of any sane human being. I get it. And I’m sure you can relate: Managers procrastinate, then rush when HR sends a reminder, and the result is that they end up with shallow, late feedback on half-remembered activities from months ago.

In the performance module we build reusable outputs: goal scorecards, feedback templates, review scripts, and calibration language.

This helps because most performance problems are execution problems.

What we explore during this training is how to create strong drafts for performance reviews in a way that reduces chaos - with HR creating scripts and templates they can provide managers to make their life easier while improving coherence through the organization.

So yes, AI can make performance management more consistent.

But unfortunately, consistency still isn’t the same as fairness.

The failure mode: the drafts are clean - but accountability is weak

Here’s where the real pitfall is: When the draft writes itself, we start acting like the thinking did too.

This is what I mean by outsourcing judgment in performance management:

• A manager accepts AI phrasing because it sounds reasonable, then looks for examples after.

• HR can’t explain a rating beyond “based on the assessment.”

• Calibration becomes a vocabulary exercise because the evidence underneath is thin.

AI makes this easier because it can produce a complete, balanced-sounding narrative from very little.

That lowers the embarrassment threshold. Reviews look finished even when the case behind them is weak.

And when someone challenges a rating, HR has less to stand on.

Subscribe now

Calibration language helps, until it becomes the work

Standard calibration language is useful, because it reduces confusion, and helps managers speak the same language.

But, and this has little to do with AI, the risk is that labels become the focus.

People start debating whether someone is “high ownership” instead of whether they 1. delivered outcomes, 2. how they delivered them, and 3. what trade-offs they made.

AI can generate polished calibration statements for every rating bucket.

That can make everyone sound equally thoughtful, but it can also hide shallow evaluation.

The part that matters: the conversation

Performance management lives or dies in the conversation, not the document.

AI can draft a review script. Helpful.
The hard part remains: saying something clear to another human being, with consequences.

I also see managers trying to use AI as a buffer. They hide behind the script, blame “the process,” and avoid the meaning.

HR ends up with tidy documentation and frustrated employees.

Share

A few practices that keep capability alive

These are small habits that protect judgment while still using AI for speed.

Keep a running evidence file all year.

A lightweight habit: managers capture short, dated examples when something meaningful happens. One line is enough.

Now this sounds cool, but making this happen requires a bit of effort. Some of my HR clients are keeping this simple, by sending a note in the system people already use, reviewed monthly for two minutes.

Then AI can synthesize without inventing substance.

This approach helps countering the disasters of year-end reviews built on recency bias and mood.

Let AI build the structure, then require humans to supply the examples.

Use AI for categories and wording options, then pause and force the manager to attach real moments before the review is marked as done.

This is done just by establishing checkpoints in the workflow where the draft remains incomplete until managers insert examples.

[Cheeky tip: Set a minimum wordcount that forces managers to actually write relevant stuff. A small number is what you’re looking for because it forces clarity and deters from including fluff]

By doing this, some of the companies I work with tell me that it has become much easier to defend feedback - especially when someone asks ‘When did I do that?’

Use calibration language, but anchor it to decisions and outcomes.

Standard language works when it points to work and impact. It becomes toxic when it becomes a personality label.

if you tie calibration statements to results, observed behaviors and context, with HR leading the editing work, AI can provide some very powerful drafting support.

The result can be very positive because it anchors language to company standards, and generates rating s that feel a bit less arbitrary.

Make managers speak like humans

AI can help structure a conversation, but managers still need to speak like humans.

Here comes another habit that HR needs to really work hard to embed in the workflow: Managers keep the flow as it is provided by HR, but must rewrite key lines in their own words before the meeting.

This should help reducing the risk [and bad feelings] of employees leaving the room feeling managed by a template.

Performance management is where HR’s credibility becomes visible.

It becomes visible because finally HR can use tools to simplify managers’ lives, instead of asking for tedious, annoying, and out-of-touch paperwork.

AI can reduce admin load and improve consistency, which is part of the intent of this training. And luckily, the risk is predictable: faster documents and weaker decisions.

Up next: development is where it becomes personal

In the next article, I’ll move to the third module: development.

Because once AI starts shaping performance narratives, it also starts shaping who gets opportunities, who gets coached, and who gets left behind - and we can now ensure that development is no longer requiring an exclusive decision between personalization and standardization.

What is slightly more unusual, is that all these sessions revolve around the hottest topic of our times: Artificial Intelligence [of course].

So, last week, I was in a training room in Shanghai with a group of HR people, watching something oddly human happen.

I give them five CVs.

Same role, same context.

I ask them to pick a top candidate and write down why.

They look at the documents, and they make their call, noting it down on a piece of paper.

Then I ask them to run the same five CVs through AI for a first-pass screen.

The room goes quiet.

I can sense the somewhat trepidant feeling in the participants. They are all very eager to see whether the tool is about to change their workload and their liability, at the same time.

The top pick comes back different.

It looks confident, as AI does.

And then someone asks the question that always matters more than the output:

Why did it pick that one?

That question is the start of this mini-series.

Over the next three articles, I’ll use the structure of a training I designed for HR teams in China: Recruitment, Performance Management, Learning and Development.

Three modules, three areas where AI can save time and also absolutely wreck capability if teams treat it as an outsourcing vendor rather than a strategic companion.

This first piece is about recruitment.

Of course, we start from recruitment because in the employee lifecycle is where everything begins.

But we also start from here, because in recruitment, speed is often highly rewarded - sometimes blindly.

And now, we have a machine that seems to be designed specifically to manufacture speed. And in doing so, it makes it look like competence.

Subscribe now

A brave new world?

General sentiment in China about the economy is, as it normally is in China, very contradictory.

On one hand, there’s an optimistic hype related to how new technologies are going to impact this ever-changing place, and the people who live in it.

On the other hand, uncertainty, a substantial slowdown of consumption and therefore, a perceived slowdown of the economy as a whole, make people uneasy.

This is China.
A place of paradox, a place of contradictions.

In this context, a lot of HR teams in China are under pressure to 1] move faster, 2] increase consistency, and 3] protect fairness all together.

It’s a rough trio.

To make this even more challenging, the Chinese talent market at the moment swings between “Oh my goodness, we can’t find anyone” and “We have 300 applicants by lunch”.

Throw in increasing need for roles where cross-border dynamics [inbound or outbound], language skills, stakeholder management, and ambiguity tolerance are becoming just as important as technical skills.

You can imagine how in this environment, AI lands in recruiting like rain after a prolonged drought.

It sounds like a miracle!

In minutes, you can finally get job descriptions done for an entire department, write interview questions, shoot out super-professional, bilingual candidate emails. And if you have received those 300 applications by lunchtime, now you can screen them before the afternoon coffee break.

Absolutely fantastic.

And even though I consider myself a skeptic, I am on the team of those who think that alll of the above is just great.

In the course I designed, we frame the recruitment module around exactly those practical outputs: high-impact JD templates, role-specific interview guides, candidate screening criteria, candidate communications.

We created this training so that HR can walk away not just with a more solid knowledge of AI, but also with stuff they can reuse and defend without reinventing the wheel over and over again.

Now, here is where I draw the line. That defend part is where things fall apart in most of the use cases I see every day in HR teams.

[And to be fair, beyond HR too - but I shall not allow myself a digression yet]

Recruitment is not an admin process.

Recruitment is an organizational decision with a [very] long tail.

Hiring mistakes hang around in the long run. Ask anyone who hired the wrong person, and they will inevitably bring up performance issues, frustration, never-ending arguments around culture fit, and in the worst cases, someone will even mention a slow decline of team standards. Ugh,

So here’s what happens past the line I draw.

AI can [and should] accelerate hiring.
But AI can also [and will] accelerate mis-hiring.

Let’s go back to those CVs in Shanghai.

Leave a comment

When the tool fills the blanks

Participants were all looking at their screens, where AI had just produced its sentence.

The question [Why did it pick that one?] hangs in the air.

We look at the screening conducted by AI.

The top profile had a great, clear story: Certifications, known employers, and a tidy career progression.

Now, the problem was, that hte participants understood immediately what was missing.

The AI selection did not account for the context of the role.

That role required someone who could connect a China team with HQ. Translation, stakeholder management, context switching, political sensitivity, tolerance for ambiguity.

That someone needed to be, first and foremost, a bridge.

We hadn’t given the AI that context.

So it did what tools do.
It filled the blanks with patterns it has seen before, and leaned on best practices and industry standards that can be read easily.

The output was solid - and it was also presented very nicely, I’ll admit.
It felt coherent, well thought-through.

Now the problem is, we were in a training. A safe space where timing is deliberately designed to spark reflection and accommodate exchange, brainstorming, and peer discussions. At least, my trainings look like this.

But real life is different. In real life, while you’re going through CV screening, you’re being bombarded with other emails, you wonder if you closed your apartment door when you left this morning, and suddenly realize that you forgot to buy groceries for dinner.

So, in real life, AI seems designed to really make us think: Oh well, it looks good enough. I can roll with this. My boss needs this by end of day.

The people in the room had time, were prompted to review the output, and they were senior and experienced enough to spot the mistake.

A more inexperienced person would not have had the intuition. A senior person under pressure would probably miss it too.

This is where the dangerous outsourcing of judgement starts.

And before you know it, it becomes a habit. And habits have this terrible tendency to compound.

For some reason, the bad ones do that better than the good ones.
[I am writing this as I realize I just made myself a third round of coffee before 11.00 AM].

A short-term win with a long-term bill

Outsourcing judgment to AI works beautifully in the short term.

In HR and recruitment, the short-term gains are obvious: Clear up inbox, delivery quickly, look responsive and efficient. What’s there not to love about all this?

The problem is, when the bill is eventually presented.

In one recent instance a client reported to me, an HR team couldn’t explain why a candidate was shortlisted.
But the real tragedy happened when one of the team members, pressed for an answer, blurted out:

The tool recommended it.

Not wise, I am afraid.

And this is just one example.

I have seen HR having to explain inconsistent decisions that were, indeed, fast. But when corrections were needed, no one really knew where to start from. Because they didn’t make them - the tool did.

I was also in a meeting where an argument about fairness couldn’t be managed by the HR in the room, because they didn’t own the logic behind the decision now being discussed.

I know, this may sound a bit over-dramatic.

But the problem is not just ethical or moral. In fact, I think the problem is operational.
[OK I think it’s also ethical, but I won’t go deep into this for now]

I recently posted a reflection on the importance of apprenticeship, and recruitment is one of its many aspects inside HR.

When you work in recruitment, you know this is where people learn how roles work, how companies compensate for missing capabilities, and what success looks like in practice.

But if we allow AI to take the first pass every time, then HR becomes a handler of outputs.

And we end up in a situation where the handlers get faster, sure. But builders get better.

And in the long run, it’s who gets better who remains.

And it’s up to us to decide whether AI gets better, or whether we do.

What does a strategic companion look like in recruitment?

When I say strategic companion, I’m talking about a specific working relationship.

In this relationship, we should allow AI to do what it is great at: First drafts, structure, breadth are the first things that come to mind.

But then, what’s the role of HR? It boils down to ownership and responsibility - of context, trade-offs, risks, and accountability.

Recruitment is full of trade-offs:

• speed vs diligence

• consistency vs nuance

• potential vs proof

• credentials vs trajectory

• “this person can do the job” vs “this person will survive the team”

If you allow it to do so, your AI companion will be more than happy to produce answers to these questions.

But don’t be fooled: It still won’t understand them.

I am lucky enough to be working with a few great HR teams here in China. What they’re doing is both reasonable, and strategic.

They are designing the workflow so that AI is forced to work inside their thinking, rather than replacing it.

Instead of treating AI as an oracle, they’re partnering with it as they would with a very fast colleague who needs a proper brief, and a proper review.

Share

A few things worth trying

When I work with my HR clients, I like to frame my ideas as suggestions rather than rules.

And I’m taking the liberty of sharing a few of them here. I like them because they create friction, but they create it in the right places.

1. Write the role reality before you generate anything.

A brief description, two paragraphs, written by HR with the hiring manager. What the person will actually do, not what the JD claims.

The point here is to try and capture the context AI will not be able to see.

If you do not make sure you do this, AI will look at a JD, and default to a fantasy job, based on generic, likely irrelevant, best practices or industry standards.

So get a quick meeting with the hiring manager, pin down the reality of the role you’re helping them fill, and then - only then - feed this information to AI. If you have a JD, feed it to AI too. Now, you have given Ai enough context to generate screening criteria, interview guides and a few other things.

That you still need to check before they are used, of course.

2. Keep a human first pass for the shortlist, even if it’s rough.

Have HR or the recruiter mark the top candidates quickly, with imperfect notes, before running the AI screen.

Then compare. Where does AI disagree, and why.

In practice it looks like forcing yourself to have an opinion early, then letting the tool challenge it.

It prevents the disastrous scenario where HR loses the habit of forming a defensible view and starts managing output instead.

3. Turn screening criteria into a decision log.

When you define criteria with AI, you also force the team to write down what they will not overweight.

For example: brand-name employers, overly polished English, certain types of certifications.

This activity is a two-birds-with-a-stone scenario. On one hand, you are telling AI not to fall into shallow selection pitfalls. On the other hand, you and the team will be ensuring a practice of critical thinking and deliberate planning.

By doing this, you’re making sure that convenient, shiny proxies do not become inconvenient and risky bias in the selection process.

[Note: this activity is something many recruitment teams I know had to establish even before AI came into play - defaulting to quick, credentials-first selection is an unfortunate by-product of efficiency at any cost]

4. Consider the candidate experience.

Using AI to draft candidate communications? Awesome.

But please, before sending those out: Read them as if you were the candidate. It takes half a minute.

If it sounds cold, vague, or dismissive: Rewrite it.

Establish this as a practice and you’ll be training your team [and yourself] not to send out stuff just because it is grammatically correct.

You’ll make also sure that while you scale communications, you don’t accidentally scale disrespect.

Subscribe now

Back to that Shanghai room

After we dissected the AI output line by line, one participant said something I liked because it was blunt and slightly annoyed.

So this is good, but it’s good at the wrong job.

Exactly.

The tool did what it was asked to do. We were the ones who failed to brief it properly, then almost failed again by treating the output as a verdict.

That’s the pattern I’m watching HR in China learn to break.

Use AI to move faster, sure.
Use it to raise consistency, yes.
Use it to reduce admin load, please.

But, for the love of all that’s good: Keep the capability, your judgement, the context, the willingness to question a great-looking output.

Resist the temptation of letting efficiency guide your work, at the expense of learning, growing, and critically assess needs, context, and goals.

AI cannot do that, unless you work yourself out of your job.

Next article, I’ll move to performance management, where the temptation to outsource gets even stronger because everyone is tired and nobody wants another calibration meeting.

In the age of fast, social media-enraged news cycles, just a few days later we all saw the explosion/implosion of Anthropic’s relationship with the Pentagon and the remark I’m talking about was soon forgotten. As it’s the case for almost anything these days, unfortunately.

So I’ll be old-fashioned, and comment on something that is roughyl 3 weeks old now. How Jurassic.

His warning, roughly, was that AI could wipe out about half of entry-level white-collar jobs within a few years.

Of course, my first thought was about job displacement, unemployment, and the potentially devastating social impact of AI.

But then, I found myself wondering about a secondary and, I believe no less important question:

If the entry-level layer shrinks, where does learning go?

We tend to think about entry-level jobs as grunt work.
[More often than I’d like to admit, I notice how in some corporations entry-level = cheap labor]

The matter here is that the entry-level is, in my opinion, a crucial part of the system where adults become professionals.

It happens slowly. it takes repetition, and mistakes that, when corrected, become learning.

So you can imagine how deeply disturbed I was, after reading about Anthropic CEO’s brutal statement.

We are enthusiastically building tools that eat away at exactly that layer.

Subscribe now

The intern summary problem

For decades, interns and juniors have been the human version of:

Summarize these documents for me.

If you’re smiling now, it probably means you had to go through it too: A meeting is coming up, your boss needs you to read a few reports, pull out the themes, and then make them digestible for someone who has twelve minutes to go through them and needs to make sense of them and look like they know what they’re talking about.

This is the type of work that lots of junior professionals end up doing. Often, we call it delegation, but it’s in fact just a way to dump boredom downward.

But it’s also apprenticeship.

When I was spending most of my time doing exactly this, I wasn’t the happiest junior consultant in Shanghai [quite the opposite, in fact].

But if I look back, I realize that while I was reading through client reports, survey results, and charts, so that I could produce information that was relevant and useful, I was being trained.

Because I was figuring out what information was actually useful, and what was irrelevant - to the final user of the information, to the context, to the goals of whatever project they were a part of, etc.

I was learning what was evidence and what was inference, and I was slowly grasping the importance of not confusing formatting with truth.

This is why entry-level jobs are important.

Learning comes from the act of doing the work, not from possessing the final output.

But now, the CEO of an AI company tells us, that AI will soon be able to do that for us.

Countless jobs lost are a problem for societies. But, I am afraid, are a temptation for businesses.

Putting [reluctantly] aside the profit-over-people issue, today I want to pause and think about the saving-over-learning issue.

This, should make the temptation fade away in any savvy business leader.

But, I’m afraid, the number of such leaders is shrinking in favor of convenience and expediency. Fast returns. Diminishing returns, too.

Subscribe now

The junior HR analyst as a canary

Let’s use one role as a proxy for many others: the junior HR analyst.
{I’ll pin everything on this poor fellow because… well, because it was me a few - too many - years ago]

In theory, their job is simple.

It normally flows like this:
1. Aggregate people data.
2. Spot patterns.
3. Summarise trends.
[3.1. Hope you got them right]
4. Support decision-making.

AI is very good at the visible part of that job.
It can aggregate, summarize, visualize, and propose interpretations with impressive confidence.

On the other hand the junior analyst does that, and something more.

While working through the example process above, the junior analyst learns to trace the story back to the data, the definitions, the timing, and the incentives.

Then, as they need to do a bt of sense-making, they learn to ask uncomfortable questions [politely].

After that, they slowly learn the difference between a metric and an explanation.

Now AI can go through the exact same steps, in a fraction of the time.
But if you let it take over the first pass every time, the junior becomes a handler of output, not a builder of judgement.

That is where capability thins.

Leave a comment

Back to the present day

I want to bring the junior analyst example one step forward - drawing from a recent experience.

A couple of weeks ago I was in Shanghai, facilitating a training with HR professionals.
We were working on recruitment processes using AI.

[I designed an “AI for HR” training that covers 3 modules - Recruitment, Performance, and Development - and makes participants use pre-tested prompt templates across a number of crucial HR activities through the HR cycle]

I gave participants five sample resumes and asked them to pick their top candidate and write down why.

Then I asked them to use AI to do a screening of the same five resumes.

The results were quite different, and the participants were very surprised.

At first, they defaulted into questioning their initial assessments. And in some cases, they were right: AI spotted things they underestimated and, on second thought, AI was making sense.

But in other cases, the AI strongly favored credentials and certifications, while the participants had initially favored a different profile, based on what they knew [or expertly inferred] about fit, trajectory, and what the role would actually require day-to-day.

Someone said, “Ok, but why did it pick that one?”

So we went through the AI output, line by line, and something became obvious.

The role we were screening for needed one critical ability: connecting the China team with a headquarters counterpart. That “bridge” work tends to be messy. It involves translation, stakeholder management, context-switching, and a certain tolerance for ambiguity and politics.

None of that context had been provided to the tool, and the tool did what tools do.

It made judgement calls with the information it had, and it leaned on what it could measure or infer cleanly. Credentials and certifications were the easiest way to assess a candidate for a job.

And, as I said, sometimes it provided good pointers.

The problem lies in the times it did not - but it was wrong while looking convicningly right.

The participants could spot the gap because they’ve done this work long enough to have internalized what matters in roles like this, especially in cross-border setups.

Even the experienced participants struggled to defend the AI choice without doing extra thinking after the fact.

The output looked solid. It was well-structured. It sounded like it knew what it was doing. The reasoning was coherent within its own little universe.

As we discussed this in the workshop, we all realized and agreed that a more junior person might have accepted it as “best practice,” forwarded it, and moved on.

That’s the learning risk, right there, in one small exercise with expert professionals.

They were about to fall into the trap themselves, and only their professional experience - and a cheeky Italian consultant - allowed them to realize they were about to outsource their judgement to a tool.

Share

The cost of speed

Organizations love speed because speed looks like competence.

[Believe me, speed is a factor that as a consultant I always have to fight even with my team. Taking 4 weeks for an organizational diagnostic is commercially hard to sell to any client. But if you have ever done one for more than 200 people, you know that 4 weeks is not an insane amount of time if you want to make sure the analysis is solid.]

Artificial Intelligence even in its most basic, commercial, non-premium form can help us reply to an email anywhere between 30 seconds and 5 minutes, give us a summary in the blink of an eye, or give us a performance review conversation script in just a few prompts.

It feels modern and efficient, and the feeling that we’re getting rid of the tedious stuff is real. Because we are.

The problem is that many of those slow, tedious parts are where people build the internal wiring that later protects the organisation.

Entry-level work teaches:

• Taste. The ability to tell good from mediocre, even when both look great.

• Context discipline. The habit of asking “what am I missing” before concluding.

• Error detection. Knowing where mistakes tend to hide, because you’ve made them yourself.

• Accountability. The ability to explain why you chose something over something else.

When AI does the early work, juniors get fewer chances to build those muscles.

And because the output often looks clean, the skill gap stays invisible until it’s too late.

These days I hear a lot of talking about skills, re-skilling/up-skilling, all the way to skill-based organizations. All good stuff, don’t get me wrong.

But if you are allowing entry level people to outsource their thinking - or if you are outsourcing human work to Artificial Intelligence for the sake of cost saving, I am afraid you are not just looking at a “junior” problem.

You’re looking at a system design problem that will soon start eroding your company’s talent pool.

I wish the issues could stop here. But there’s more.

In the next segment, I’ll be trying to answer a question I keep asking myself: is AI going to standardize us to death?

Leave a comment

What gets standardized

AI output tends to converge toward common patterns, common phrasing, common assumptions.

And this is not inherently bad. In HR, for example, some standardization is healthy.
[Compliance, documentation, consistency, all necessary]

But if the entry-level layer stops learning the craft, HR risks becoming a function that increasingly runs on:

• generic templates,

• generic interpretations,

• generic judgement.

And this is how we will end up with HR teams that produce lovely reports and artefacts, but have no idea how to handle edge cases.

And edge cases are, in fact, why HR exists.

Just like the candidate screening example: Certifications will not tell you if the candidate you’re selecting will be a fit not in the role they applied to - but in the next one.

What will tell you that is the contextual knowledge you have of the company, the current people in the candidate will report to [what’s their style? Is there match with the candidate skills, abilities, and knowledge?]

Cross-border roles.
Politically sensitive moves.
Managers who do not say what they mean.
Employees who are technically fine and socially toxic.
Performance issues camouflaged as “culture fit.”

This is where experienced HR people earn their keep, not through the document itself, but through the choices behind it.

If we gut the apprenticeship layer, we increase the probability that HR becomes administratively impressive but substantively irrelevant.

And when that happens, everybody loses. The company, the people, the market - I don’t want to sound dramatic, but I reckon society will lose, too.

Subscribe now

What will keep learning alive

While I am still trying to figure out the one big thing we can do to make sure we don’t outsource our junior colleagues’ brains to AI [when I get there, I’ll trademark it and be the billionaire I am not planning to be], here are a few things I am recommending my clients.

I’d love to hear your thoughts on them - and if you have other recommendations, leave a comment! I’d be really happy to exchange ideas.
Here’s a convenient button for you:

Leave a comment

Idea #1 - Keep the “human first pass” for learning roles

As a leader, establish a routine with your junior team members.

1. Ask them for raw drafts of the narrative first - promising no judgement on format and language.

2. Provide feedback on content, direction, connections - this is where your seniority will provide context, understanding, higher-level strategic/critical thinking.

3. Allow the junior to now use AI to see what comes out.

4. Compare their draft with the tool’s output, and reconcile differences.

By doing this, you’ll ensure that the “judgement muscle” is trained and not atrophied by AI.

AI becomes a companion, not a competitor for brain usage.

You end up with juniors who can not only defend their analysis [or whatever the output of their brainwork should be], but who are now enhancing them with this incredible tool we now have at our disposal.

Idea #2 - Force context into the process, on purpose

Engage your junior team member in a brief context-setting meeting before they use AI.

Let them brainstorm with you or other senior colleagues, what the context of an activity actually is. Let them draft a context briefing they can provide to AI.

by doing this, the AI output will be relevant and more defensible - but the team has had time to ensure they are clear on what the context is.

This should reduce the blind acceptance of great-looking, but painfully dumb and often risky AI output.

Idea #3 - Take AI output apart

This happens after a junor has produced something with AI. Make it fun: Sit down at a table, and together with your trainee, look at the AI output and question it.

What is trite, irrelevant AI slop? What is just dumb compliance with web-searched best practices? Does this actually fit your company? How is this a potential mismatch with what you’re looking for? what coudl be the unintended consequences of the AI-generated output?

This exercise helps in many ways: First, it’s a great mentoring session where the trainee has a chance to get tremendous amount of contextual and experience-based insights from their seniors. Second, it establishes a routine of questioning AI rather than just rolling mindlessly with it.

Idea #4 - Do it the ‘old way’, as a recurring exercise

You may be too busy to do the things I have suggested so far. I get it.

So why don’t at least ensure a cycle where, once a month, juniors are assigned a set of tasks that must be done manually, with feedback.

You do it the old way, and you make it a special event instead of a daily practice.

Obviously, this is a compromise, but I argue - a compromise will always be better than nothing.

By doing this, you are still building opportunities for human work, pairing them with invaluable feedback on the activities’ output.

Share

There’s still more we need to figure out.

As I re-read the draft for this article, I realize that I don’t have a clean answer for what replaces entry-level apprenticeship if entry-level jobs shrink quickly.

I also don’t trust the comforting idea that “we’ll train them differently” will magically happen at scale.

The problem is that, I am not seeing yet anyone who knows what differently actually looks like.

Because we are all learning what AI is, let alone what it actually can do. Boy, even the owners and creators of AI seem to be quite unsure of what they’re actually creating.

We’re seeing the Modern Prometheus taking shape right in front of us. I can only hope it won’t ask us for a bride.

I digress.

The problem is that right now, most organizations are already underinvesting in apprenticeship when it’s cheap and obvious.

I can’t imagine this trend will become any better moving forward.

So yes, maybe Amodei is right about the scale of displacement.

The question I keep coming back to is simpler and more operational.

If the bottom layer disappears, where do future seniors come from?

Right now, many companies are chasing speed and efficiency and calling it progress. That’s understandable, of course, but that’s also a clear, frightening trade-off that we all seem a bit too eager to be OK with.

And I suspect we’re going to feel that trade a few years from now: rooms full of adults who can produce outputs quickly, and fewer people who can explain, confidently, how they arrived there.

I keep seeing it show up in small events, almost incidents, like that resume exercise, where the output looked excellent and the reasoning went missing.

That’s the kind of problem that stays invisible right up until you need judgement under pressure.

And with the way the world is going, I can’t stop thinking how fast we’ll be reaching that moment.

Share Margins & Meetings

It feels like last summer’s China e-commerce war came back this February, with a new twist: AI assistants placing orders for you, at scale, with red-envelope subsidies behind them.

And unsurprisingly, it became a big, fat mess.

Systems buckled - and so did the people

The core mechanic is quite simple, and the script may look familiar to any China-watcher (or China resident).

Tech giants throw money into the system during the Spring Festival marketing window.
Users get red envelopes, vouchers, discounts.
Platforms spike downloads, daily active users, and whatever internal KPI makes the weekly review feel less scary.

This year, they wired the subsidies directly into AI tools.

You ask the chatbot, it orders for you.

Alibaba’s Qwen campaign pushed vouchers for drinks and processed millions of orders in hours, according to multiple reports.

As shops got flooded with instant orders, many unclaimed, many arriving all at once, systems buckled.

The problem is, people buckled too.

I’ve been reading accounts of receipt printers “churning out long strips” of unclaimed orders and staff fielding nonstop questions from delivery riders who didn’t know what was real, what was duplicated, what had been paid for, what had been cancelled.

Now, here’s the thing: High-level data does not tell you the full story.
App-store rankings skyrocketed for all delivery apps.

Somewhere at Alibaba, KPIs were being met with, I can imagine, great satisfaction.

AI did its job: efficiently, with machine confidence, and its usual machine speed.

But meanwhile, humans in the field had to face the consequences.

Subscribe now

Who pays for this?

In one of the many articles and reports circulating on Chinese social media, I read this story of an 18-year-old part-time worker at a milk tea chain describing cups stacked “taller than adults,” and deciding to drink less water to avoid bathroom breaks.

She quit the next day.

I don’t know her, and I’m not going to pretend I do. Unfortunately, I also don’t need to. Anyone who has spent time around frontline service work in China knows how brutal it can be.

Now add a “festival treat” campaign designed to create a surge in traffic, requests, and consumption.

It creates a specific kind of cruelty, because while the demand spike is artificial, the stress is very, very real.

A lot of the orders are cheap or even free to the customer, subsidized by someone else’s budget, and paid in human cortisol by workers who have zero voice in the campaign design.

On social media, this is often perceived as something silly - with funny videos of chaos in stores going viral.
In the worst cases, consumers complain about their orders being displaced or cancelled.

What does not become viral on social media, is that while this chaos is unravelling, someone still has to clean the counter, explain to angry riders, deal with refunds, deal with cancellations, take the blame from a hyperventilating manager who is also stuck in the same storm.

I normally try to stay somewhat balanced, but I have to say, pushing consumption through subsidies during a broader slowdown, mainly to chase growth optics, really looks like blind pursuit of growth optics rather than innovation.

I don’t think it actually helps creating demand. And the human cost feels way too high.

The short-circuit here is that these subsidies and related chaos impact mostly a critical segment of the population: Youth.
Just like the 18-years-old employee at the milk chain.

Chinese youth is faced with a transition period where the promises they grew up with [with the equation work hard = get rewards] are no longer being met. You might be familiar by now with the concepts of involution and lying flat, which we discussed extensively just a few months ago.

And although this is not the only reason why Chinese economy is slowing down, youth unemployment, uncertainty, and the resulting disillusion play a role in it.

Crushing this same youth with some corporate-subsidized martyrdom does not sound like a great strategy.

[I suspect, there will soon be another scolding for the Chinese delivery giants coming from Zhongnanhai]

Share

Big dreams, big words, big troubles

If you read the more official coverage, the language is grand.

On Chinese media, this is often - and unfortunately - framed as a sort of race for the next AI gateway.
A new front door to commerce, travel bookings, entertainment, payments. AI agents doing end-to-end tasks. Apps fading into the background.

I [somewhat] get the strategic logic. Whoever owns the interface owns the traffic, and traffic becomes bargaining power.

But campaigns like this also reveal a basic operational truth: you can’t “agentic commerce” your way out of capacity constraints.

You still have:

• stores with limited staff and equipment

• riders with limited time and physical stamina, and even less safety and job security

• customer service that becomes the complaint sponge

• brittle systems that behave fine, until they come crashing down

These days, as I work more and more with artificial intelligence [and while, like many others, I try to make sense of it], I have this increasing feeling that the tech narrative treats the world as a clean workflow.

In my line of business, it’s a bit like HR service providers offering AI workflows that are not really designed for flexibility, just industry benchmarks. A big problem if you ask me - because industry standards don’t tell you what the unique reality of an organization truly is.

The real world has kitchen space, queue space, printer paper, and people who need to pee.

[Yes, I know. Poetic.]

Subscribe now

A song of fear and greed

These subsidy wars are often framed as competition, innovation, consumer benefit. To be fair - they can be, sometimes. But they also smell like panic.

To put it even more bluntly, they sometimes feel like a show of fear and greed.

Fear because profits seem to be slowing down, and businesses immediately default to focusing on achieving and possibly exceeding financial targets to keep shareholders happy, rather than thinking about how to more sustainably achieve different, alternative goals.
And with AI seemingly powering up opportunities for awesome-looking spreadsheets and financial reports, it’s no surprise everyone is jumping onto it.

Greed because if you win the AI gateway, you can tax the ecosystem later. Platforms have done this before.
The difference is that the front line has become even more disposable.

And let’s be honest about the labor dynamic.

When the campaign succeeds, leadership celebrates adoption.

When the campaign breaks operations, the pain gets pushed downward.
Teams on the frontline should have prepared. Employees should be more efficient. Customer service should handle the volume. Someone is always not resilient enough. [Ugh]

This is how you teach an industry that human limits are a rounding error.

I work with organizations, not platforms, but the pattern is familiar: aggressive targets paired with magical thinking about capacity.
Then shock, horror, and surprise when burnout spikes and attrition rates go up.
Then, the inevitable meetings about engagement.

Share

Two things can be true

AI-enhanced ordering is useful.

To name a few things that come to my mind almost right away:
It reduces friction for users. It can be great for accessibility. It can help people compare options and execute quickly. It can help small merchants if the flow is predictable.

But the version we just saw was not designed for this.
It was designed for impact. The goal was to get a nice-looking spike on a chart, possibly a PR-boosting headline and a top 10 app-store ranking.

There’s also a weird cultural side effect: when the experience is chaotic, people start distrusting the AI itself, even if the underlying model is fine. Users remember the day the chatbot “worked” and their drink never showed up.

As a result of this chaos, Alibaba’s Qwen had to pause or limit parts of the coupon giveaway and asked shoppers for patience after the system got overloaded.

Now, patience is a funny word here.

The system asked humans to absorb the volatility created by the system that was supposedly designed to make their experience better.

Ironic.

And also telling us that maybe, just maybe, humans did not figure prominently when the idea was floated and then approved in a corner office.

Subscribe now

From milk tea to management

It would be comforting to treat this as a weird China e-commerce anecdote.

But this is a pattern we should learn something from. I think.

Just try and swap AI order placing with any internal AI rollout and you get the same script. Someone removes friction, demand surges, and the system downstream takes the hit.

Only inside a company, the downstream is your operations teams, your customer support, your finance controllers, your line managers, your employees doing workarounds in silence.

AI makes things faster. It also makes consequences arrive faster.

So when AI pops up in a leadership meeting as the shiny shortcut for efficiency, impact, profitability, workflow speed, whatever the word of the week is, I recommend you keep some guardrails up.

By no means, these are not to be intended as precepts carved into stone - but a few useful things to keep in mind before you roll out something that is bound to do more harm than good.

1) Speed creates debt

In this story, the chatbot removed friction, so volume hit the counters like a wave, while workflows and capavity couldn’t meet this new speed.

Before you roll out an AI tool, name the new peak it will create. And then ask yourself the real question: Who carries that peak?

If your AI reduces cycle time by 30%, where does the extra throughput go?
Approvals? Customer responses? Production? HR’s inbox?

Time saved is a good metric to evaluate operational efficiency, I get it. But not if it shifts the work into bottlenecks, overtime, and burnout.

In the promo story, the saved time resulted in absolute chaos in the field.

I have seen something similar the other day, where an AI tool for performance management - designed to expedite and free up time for managers who could now draw performance information for their people from emails, work logs and some other sources - resulted in disgraceful performance conversations that needed to be re-done because… well, because managers were not prepared to have them.

They did not really think about them because they didn’t have to. And then realized that they couldn’t talk about them with people who, rightly, wanted explanations and not just AI scores.

2) If the business case works only because someone absorbs pain, it’s a bad business case

Subsidies benefited users and platforms. Labor absorbed the shock for free.

My recommendation is, include workforce impact in the ROI.

Headcount, training hours, exception handling, QA time, escalation load, and the cost of churn when the people closest to the work get sick of being the shock absorber.

The pursue of efficiency should not become a mere externalization of badly designed workflows onto those teams and people with the least power to psh back.

Ask any IT leader who ever had to work on automation. They’ll tell you how difficult it is to make sure that automation pushed from the top does not equate into long-term disruption and messy, unclear, and even poor results in the front line.

We’ve seen this for decades, yet now with AI it looks like we never had to tackle this exact issue.

3) Exceptions are the product

In the story oif AI-boosted orders, the disaster took the form of unclaimed orders, duplicates, cancellations.

So here’s what companies must do when they’re preparing any type of AI enhancement: Design the exception path right away.

Focusing on what will work is the big part of this, of course, but anticipating what can [and most likely will] go wrong is just as crucial.

Who reviews outputs? Who overrides? What happens when the tool is wrong, or confidently wrong? How does a frontline employee flag it without starting a political fight?

We’ve been playing around with AI tools for less than two years now, and I’ve seen this happening so many times already: AI rollout pilots that look just fantastic, and then turning toxic when deployed at scale because nobody owned the “when things go weird” situation.

4) Don’t roll out AI until you can explain it in one sentence

Users didn’t understand what was happening, and riders didn’t either. Confusion filled the gaps.

Leaders should be able to answer, plainly: What does the AI do, what does it not do, and who is accountable for decisions? Then repeat that sentence until people stop improvising their own version in WeChat groups or Slack channels.

If leaders do not do this, adoption becomes rumor-driven.

Because of the work I do, I have plenty of experience in this type of rumors when it comes to AI tools being adopted at various points of the HR/employee lifecycle.

When communication is not crystal clear and transparent, people will inevitably start seeing the new, flashy AI tool as a symbol of cost cutting, surveillance, or management laziness, even when the intent was decent.

5) Pilot for human behavior, not for model performance

In today’s story, the system worked! But only technically.

Humans failed to cope with the surge and ambiguity.

Pilots are often thought as a mean to test the new system, but out of fear, they’re often roll out in ‘safe’ environments - to avoid problems [mostly political] or excessive disruption.

I think that’s a bad strategy.

Pilots should be rolled out where and when real stress happens: peak volume, messy inputs, adversarial users, tired teams, managers who have other priorities, imperfect data.

Watch how people actually use it, and the shortcuts they take.

The problem here is that we are too easily confused: We look at technical readiness and mistake it for organizational readiness. The model might be fine! But it can still blow up your trust and your workflows.

Share

If I had to translate those reflections into choices a leader can actually make, it would look something like this:

• Budget for humans. Allocate time and people for training, QA, escalation, and feedback loops as part of the project plan, not as an apologetical afterthought.
  By doing this, you can reduce the risk of AI becoming a tool that works, while everyone hates it.

• Set a throttle, even internally. Phased rollouts are your best friends. Do them the way that makes the most sense for the business: By teams, by volume caps, by function. At the same time, consider the keen beans: Set up opt-in windows early adopters who can tolerate turbulences.
  The upside of this slightly heavier up-front work is that problems come up more distributed and can be fixed as you go, and whatever new AI tool you’re rolling out is perceived as a positive improvement, albeit a work-in-progress, rather than a fiasco.

• Name an owner for the outcomes, not for the tools. Normally, IT and HR are assigned as owners of the tools. This is great because at least you know who to blame when it goes sideways.
  But now we know, this creates a stalemate because when things go wrong, it’s the business (and not IT and HR) who have to deal with the aftermath.
  Keep IT, HR or anyone else in charge of the tool accountable - but make sure that a leader [or a group of leaders] remains accountable for the expected bbusiness outcomes. Distributing ownership and accountability is the only way to make sure that the tool keeps going in the direction the business needs.

• Measure the right [ugly] things. Track rework, escalation volume, cycle-time variance, error rates, and attrition hotspots, not only adoption and cost savings.
  The least desirable thing in any organization is having leaders in townhalls celebrating the wrong metrics.
  It’s the biggest turn off: Think of a leaders trumpeting great successes based on usage ratings, while people listening know they had to do stupid, meaningless, overtime work to handle a broken system.

I keep thinking about the 18-year-old in that milk tea shop.

First job, maybe. First real rush. The kind of day that teaches you, very quickly, where you sit in the hierarchy of important. Somewhere below the coupon logic and the growth chart.

She got crushed by a system that will never remember her name.

The dashboards won’t. The postmortem slide won’t. The exec recap definitely won’t. Someone will call it a learning and move on.

She won’t move on that fast.

She’ll remember the printer spitting orders like a joke, the rider snapping, the customers filming, the supervisor barking, the moment she really neded a bathroom break but couldn’t leave the counter.

Then quitting the next day, because that’s what you do when you’re eighteen and you realize what’s the deal on the table.

So when AI shows up in your organization as the fast route to efficiency, keep her in mind. A real person who took the impact of someone else’s cleverness.

AI can generate speed in seconds.

The rest of the system still runs on people.

And people, inconveniently, remember.

It’s not by happenstance, of course: I’ve been working on a partnership with a company offering observation-based assessments for talent development purposes, with interviews conducted by an AI interviewer.
Their tool is not only very well designed, but sits perfectly within my team’s toolkit for strategic talent advisory to our clients. [Note: I promise, this is a Margins&Meetings reflection, not a sales pitch]

So here I was, in one of these chats, talking with the head of HR of a large organization operating across Asia.

The opening statement:

We want to revamp our 360. But the cost is insane. And the optics of AI… I don’t know.

He paused on the word optics. He didn’t love himself for using it.

Then he did what clients have to do: He walked me through the math.
One candidate interview. Up to five sponsor or stakeholder interviews.
Then the scheduling, the follow-ups, the synthesis, the readout.

You can imagine the amount of billable hours at the bottom of the spreadsheet.

The moment we finished our calculations, it was painfully obvious that the real issue was: They want 360-degree feedback because it’s useful, but they also want it to be viable.

And with traditional 360-degree feedback, viable ends up meaning: For a small number of people.

The tradeoff we should talk about more

The sad thing is that companies are forced to make a choice that is not between human-led 360s and AI-led 360s.

They’re actually choosing between a tiny, expensive 360 for the top and no serious 360 for everyone else.

The problem, the tradeoff is all about budget, whether anyone wants to name it.

So when people get anxious about an AI interviewer replacing humans, I get it. We are all a bit concerned with AI taking away our jobs. [I am!]

But I also think it sometimes distracts from another sad and much more actual reality: human-led 360s, done properly, are labor-heavy enough that they become scarce.

And scarcity is not always a good thing for organizations: Especially when people-related initiatives are involved, scarcity of investment [and opportunities] inevitably creates tiers, and people end up reading development as status.

And feedback feels like a leadership perk.

[Expert tip: Do you want your mid-level leaders become cynical monsters? Ensure only their bosses get leadership development and coaching opportunities]

But yes, 360s are now a luxury good. Let’s see why.

Subscribe now

Why 360s became a luxury good

I love 360-degree feedback surveys, but traditionally, they really do take a lot of work.

There’s the prep with the candidate. The stakeholder selection dance. The “let’s make sure we have a balanced view” conversation that can so easily turn political.

Then we’ve got a team of people going through interviews, notes, synthesis, theme-building, and careful wording so the message doesn’t detonate once the individual reports are out.

It can be excellent. I have had the opportunity of being part of some amazing 360 projects over the years.

But it can also become something heavy and over-engineered that everyone defends because it’s expensive.

There’s a strange belief in some companies that cost equals quality.

If it doesn’t hurt, it can’t be good. If it doesn’t take twelve hours of calendars and a 40-page report, it can’t be serious.

Meanwhile, who gets excluded?

Lower levels, almost by default, end up being excluded from 360s because nobody can justify the time and cost at scale.

Everyone knows that they would benefit greatly, if not most, from 360s. But there simply isn’t a way to justify the cost, it’s just too-fricking-high.

So the 360 becomes something you earn by reaching a certain grade.

And then we wonder why feedback culture never really takes root.

What the AI interviewer actually changes

This is where the AI interviewer comes in.

I’ve been working more and more with this one, and I have been very impressed by the amount and quality of the training this AI interviewer has gone through to get to a point wehere it’s able to conduct fairly impressive behavioral interviews.

Think of it as a structured interview layer that can collect and organize input at volume.

It asks questions, probes, clarifies. Just like one of my team members. And myself, too.
And it produces solid outputs. Like, really solid.

So, it’s able to conduct good interviews. Great, but that’s only 50% of the value.

The other 50% is that it reduces the most expensive part of the process: collecting structured feedback from multiple people, consistently, without turning the whole thing into calendar Tetris.

And all of a sudden, as collection becomes cheaper, you can make 360-degree feedback viable - and available to more people, not just the top.

But as I noticed in my chat with the client above, cost is not the only issue. The other problem is: Can we convince people an AI interviewer is a good thing.

Share

Optics are not a side issue

The client’s hesitation was more social than it was technical.

He didn’t ask:

Will it work?

He asked:

Will people think we’re cutting corners?

That’s a fair fear.

Many employees hear “AI” and translate it as cost cutting in the best case, or outsourcing. Worst case: We’re being experimented on.

In some companies, trust is thin enough that people can turn any new tool into a story about disrespect.

So yes, the optics matter.

And believe me, optics are important because when it comes to talent and leadership, we have f****d up so many times over the past few decades. We wonder why people are so irrational, but we should actually just realize, that people in fact are just… experienced.

The AI for talent mess

There’s also the current market noise.

AI is sprouting all over the employee lifecycle. Assessment, development plans, coaching tools, performance prompts, recruiting screens, sentiment dashboards.

Some of it is serious. Some of it is an empty void with a trendy AI label.

And too often, the tool is used as a shortcut around doing the hard internal work.

There was this one team which bought a new “AI” tool because their CEO forwarded an article and wrote “we need this.”
Two months later, HR is presenting outputs they don’t fully understand, with the vendor’s language slapped across their company slide templates.

The tool looks confident, but the team looks nervous - because they are not entirely sure they can defend the findings of a tool they don’t really understand, or have the competence to take apart.

Because they outsourced that competence.

It’s an easy trap, especially when the organization already struggles with feedback literacy.

I do believe that skepticism about AI in talent practices is healthy.

Subscribe now

Human(s), after all

I think that the AI interviewer is interesting because it protects human time for the stuff that actually require humans.

Let me give you a few examples.
A good AI interviewer can do something humans struggle to do at scale: ask the same questions with the same patience, every time.
It doesn’t rush because the next call starts in five minutes.
It doesn’t really have a bad dayy and it cannot be in a bad mood.
It doesn’t get distracted while a candidate inevitably starts rambling about something incredibly boring and irrelevant.

And that’s not all.

In my experience, especially in organizations that experience strong politics or that have a face-saving culture, there are things people won’t tell their boss or HR.
That’s where I come in as an external consultant: People does tend to tell me things they wouldn’t share internally.
And that’s the cool thing: I have already noticed how, in some contexts, people will tell a screen things they won’t tell a person.

Obviously, we’re all different.
Not everyone likes to talk to a consultant, and just like that, not everyone will feel safer talking to AI.

Either way, the value is straightforward.

If you spend less time collecting the raw material, you can spend more human attention on:

• framing the purpose so it doesn’t become a popularity contest

• checking patterns so one loud voice doesn’t dominate

• helping the candidate read it without spiraling

• coaching the line manager so this turns into a real conversation

That’s where meaning gets made.

The collection method can support that.

Of course, it can also sabotage it, if you treat the output as a final verdict.

That’s the human part we cannot, and should not, get rid of.

Scaling creates a second problem

Now, I need to make on thing clear.

Sure thing, as you probably already noticed, I am advocating for scaling and accessibility of 360s.

But I must be honest here:

If you suddenly make 360s accessible, you also suddenly create demand.

You are producing more reports, more managers are pulled into feedback conversations, and inevitably, more risk that someone will read one line and get stuck on it.

This means that accessibility and scale requires a honest look at the organization’s capacity to handle feedback.

If the company has no muscle for feedback conversations, faster collection just produces faster confusion.

I say this as someone who has, more than once, watched a well-intentioned feedback process land badly because the manager doing the follow-up had no idea what to do with it.

The report was fine, but the conversation was the disaster. [I am sure you have experienced something like this, too, either giving or receiving said feedback. I sure did.]

So the question isn’t only if we should use an AI interviewer.

It’s also what are we doing with what we collect?

Three safeguards to anticipate pain

A few suggestions I’ve been giving clients when they consider an AI interviewer for 360s.

Be painfully clear about what the AI does

People fill in the blanks with their worst interpretation when you leave space.

What do you do when someone asks who’s going to read the output of the 360 interviews?

Answer in one sentence.
You say who sees raw input, who sees themes, and what happens next. You also say what the AI is doing, in plain language, without vendor poetry.

If you do this, you are anticipating the rumor cycle where AI 360 becomes “they’re automating people decisions.”

My recommendation: Be transparent and do this, even if nobody asked out loud.

Keep a human in the loop

Collection can be automated. Interpretation still needs context.

We’re talking about AI interviews, but it truly applies to anything AI- and people-related.

With AI 360s, you must ensure that a trained person reviews themes for coherence and checks for weird outliers before anything gets shared.
If something feels off, they must be able to spot it, and investigate instead of trusting the formatting.

A trained analyst can ensure you avoid the misstep of sending out a report that looks precise but is miscalibrated, leading to mistrust or an overreaction.

Design the follow-up before you run the first interview

Output without support creates anxiety.

Launcing 360-degree feedback projects means that you have already designed the full process, not just the data gathering.

You need to make sure managers know how to run a debrief conversation. When not possible, we discuss with our clients that the debrief is in fact run by one of us.

Candidates need to be given reassurance, that their managers are reading and interpreting the same report.

We often set up brief training sesions with managers to help them understand the reports - so that they can have conversations consistently across the organization.

This prevents a big risk, of people using feedback as a verdict, or even worse using it as ammunition.

[It’s worth noticing that this last recommendation applies to any feedback system, not just the AI ones. Please, train your leaders on giving feedback.]

Back to that client call

Near the end of the call, the client came back to the optics.

People will ask why we didn’t use real interviewers.

I told him: people already know the current system excludes most of them. They may not say it out loud, but they notice who gets the developmental investment and who gets a training catalog and a vague pep talk.

We ended up talking about fairness more than technology, because if feedback remains expensive, it remains exclusive.

As I said earlier, scarcity and exclusivity shape culture in strange, often bas ways, because on one hand it limits insight, and on the other hand, it sends a clear message on who counts.

The client, off course, agreed.

At the end of the call, he stated another important factor:

I just don’t want this to look cheap.

I understood him.
Nobody wants to roll out something that gets mocked internally as HR trying to save money.

What I told him [OK maybe a bit too bluntly]: Plenty of expensive 360s still produce cheap outcomes.

We laughed. Just a little.

Then we went back to the real design questions: Who gets access, who supports the conversations.

And most importantly, who owns the competence and how do we make sure people uses the output of the 360s over time.

We discussed how to embed the results into performance conversations, talent plans and succession planning.

And this is what really matters, because the whole thing lives or dies on how it is actually used, long after the AI interviewer has asked its last question.

[I recommend you go to and read the article, it’s a truly fascinating insight into how Chinese social media actually work, and the way they can really shift business strategies in the blink of an eye]

As I was reading the article, I realized that this story might also be telling us something really interesting about employees, and how quickly affection and loyalty can turn into frustration.

So, today’s reflection is on emotional investment, from an employee perspective rather than a customer one - but you’ll see, I hope, how the two things are strikingly similar.

They are similar because companies today are increasingly relying on something very close to fandom from their people [although they may not call it that].

This Xiaomi social media fiasco is not [only] a China story in the cultural sense.

I think it goes a bit beyond that because the way I see it, China just happens to be a place where dynamics like platforms, speed, and visibility show up early and [sometimes, somewhat] clearly.

I often tell the leaders I work with, that they should look at China like a perfect environment for stress tests for trends that may soon become global.

This might be one of them.

From customers to colleagues

Xiaomi has an interesting story when it comes to their customers’ loyalty.

Since they started up, Xiaomi made some impressive efforts in focusing not merely on their product features [to be fair, until just a few years ago, Xiaomi products were not exactly top-line - but they were affordable] but instead on building a sense of participation.

People felt included and listened to, like they were part of something bigger than a transaction.

And over time, that emotional investment became part of the brand’s value.

[I cannot count the number of times, over the years, in conversations with Chinese clients, partners and friends, they identified Xiaomi as possibly their favorite Chinese tech brand - even if they were not Xiaomi users/fans themselves]

Now the problem with this level of emotional investment is that when interactions land badly, the reaction is definitely more than… polite disappointment.

It becomes personal.
Just like in this Xiaomi PR nightmare, Chinese fans tend to react as people who felt let down by a relationship they had taken seriously.

So here comes a pattern I see inside organizations, across industries and geographies.

Many companies have spent so much effort and so many years encouraging people to feel proud and connected. To feel emotionally invested.

Leaders talk about purpose and belonging, and keep ranting about shared journeys.
Teams are asked to stretch and to care more, and more deeply.

I am not opposed to this approach.

If done well, it’s a win for both the employees and their employers because it creates energy and commitment.

The side effect, however, is that this also creates expectations.

Subscribe now

The emotional contract at work

Whether leaders acknowledge it or not, every employment relationship includes an emotional layer.

People carry assumptions about fairness and their need [right] to be heard.

When those assumptions hold, tolerance for mistakes is surprisingly high.
When they break, reactions can feel sudden and sometimes even disproportionate.

This is where the parallel with customer fandom becomes useful.

Fans forgive missteps when they feel respected. Employees behave in much the same way.

When people feel dismissed or talked down to, the reaction is not very loud at first, but it can quickly turn into disengagement and cynicism.

The point is, the emotion itself is OK - but ignoring it is not.

Ignoring the emotion, or pretending it doesn’t exist, is where real risk lies.

Social media as the amplifier

One thing that has changed rapidly [and massively] over the past decade or so is visibility.

People now narrate their experiences publicly.

A frustrated customer post can go viral in hours - sometimes even less.

A disappointed employee reflection can reach future candidates well before HR sees it, or do something about it.

Obviously, not every complaint is fair or accurate. But ignoring them is no longer a viable option, regardless of their accuracy or truthfulness.

Social media are merciless and, if I may add, a bit shallow. Everywhere in the world, definitely not just in China.

And again, I am not talking just about customers.

Social platforms have become an extension of organizational culture, whether leaders like it or not.

Leadership is the real bottleneck

When backlash happens, the reflex is often to fix messaging.

PR teams rush at pumping out new guidelines and polished statements.
Of course, they matter - but they rarely address the root issue.

The deeper challenge sits with leadership behavior.

As we all know, at work even the smallest moments carry disproportionate weight.

Think of the disatrous effect of a careless joke in a town hall, or a dismissive response to a concern.

one of the most regular issues I hear from people I coach is what happened when a decision explained as obvious by a leader was, turns out, really not that obvious to the people impacted by it. [Ouch]

These moments accumulate.

And over time, they shape how people interpret intent, eventually leading to an erosion of trust. And as trust erodes, even neutral actions get read negatively.

[Please let me clarify - I say this with all the humility I can, because none of us are immune. I have seen leaders I absolutely respect underestimate how their words land. I have done it myself]

The danger of turning employees into fans

This where things get a bit tricky.

On one hand, the benefits of emotional engagement are clear and well-researched.
Employees who care tend to perform better and collaborate more effectively. As a result, employees experience higher levels of job satisfaction and companies report lower attrition rates.

At the same time, fandom raises the bar.
Emotional commitment creates a demand for reciprocity. People who are emotionally invested have clear expectations. To name just two, they expect to be heard, and they want their leaders to act in line with the values they promote.

So you can imagine what happens when companies ask for loyalty [often presented in the form of stretch assignments, performance improvement, extra time for learning and development, overtime work or even worst, team building activities on a Saturday - stuff of nightmares] but fail to reciprocate with respect, true or perceived.

When this happens, the gap becomes really, painfully visible.

None of this implies malicious intent.

In fast-moving environments, leaders are under pressure, and trade-off are not only real, but often necessary.

Unfortunately, the problem remains.

Pressure does not cancel perception.

Power, emotions, and culture

Value statements and engagement scores are but one aspect of company culture.

Another aspect of company culture is how power responds to emotions.

In practice, you see real culture emerge in the way people raise uncomfortable questions. or when mistakes happen and someone has to explain why.

The Xiaomi story highlights how fragile emotional capital can be.

It takes a few moments of distance or perceived arrogance, even unintentional, to weaken years of goodwill.

Organizations that recognize this tend to invest more in how leaders listen as well as in what they say.

They accept that emotional dynamics are part of performance, not a distraction from it.

Share

So, what.

I do not think companies should try to make employees into fans.

That path carries risks.

I do think leaders need to acknowledge that many people already relate to their workplace emotionally, whether that was the original intention or not.

Once that is accepted, we can finally shift the question.

How do we lead relationships, not just roles?

A useful place to start is paying attention to where friction actually shows up.

Surely, engagement dashboards and regular surveys are useful and interesting, but you must keep an eye on everyday behaviors.

For example: Is your people asking questions, or meetings go quiet after a decision is announced? Or, is the same concern resurfacing in different form but through different people?

These are real life examples, things I saw happening with my clients. They were labeled as attitude problems or changer fatigue, but the leaders I was working with soon realized that they were signals that something in the system was grinding.

Leaders who learn to read these signs can treat them like warning lights on a dashboard, not like personal attacks.

Another helpful discipline is slowing down communication just enough to sanity-check it.

Before a town hall or a big announcement, share the message with people who were not involved in the decision and ask one question only: how could this land on a bad day?

Not the official interpretation, the grumpy one.

A few minutes of this usually reveals where the message sounds off. Adjusting early is far cheaper than explaining yourself repeatedly after Slack, WeChat, or the internet has already filled in the blanks.

The last piece is being clear about the deal on the table.

When extra effort is needed, say so plainly and acknowledge what the compromise actually is - what people are giving up.

When stability cannot be promised, resist the urge to wrap uncertainty in motivational language.

The thing is, adults tend to handle difficult realities better than vague reassurance.

What creates resentment is often times not the pressure, but the perception that ambiguity is being deliberate, that your words are unclear because you’re hiding something.

Even if you are not, you’ll be perceived as untrustworthy, especially over time.

Clear expectations and visible trade-offs may not make leaders popular, but they do make them credible.

While none of this is complicated, I have to admit, it is also surprisingly rare.

But why is it?

The Xiaomi episode is a good reminder that strong cultures are built less on hype and more on trust.

Followership grows when leaders are clear, consistent, and human, especially in small, unglamorous moments.

People do not need to cheer for every decision, but they do need to understand it.

Fandom runs hot and cold. Followership tends to stick around.

When leaders act as stewards of the relationship rather than brand ambassadors, they don’t need slogans to engage with their people.

In part 1 we described the fog machine problem.
Big words drift into the room and everyone nods.
And then each person walks out with their own private definition of what they just heard.

That is how so many decisions in companies of all types and sizes end up depending on interpretation and confidence, and whoever tells the best story in the room.

In this reflection, we look at lighting design.

[I know, I know. This stage/drama metaphor might be a bit stretched but hey, I can’t help it]

Before we move on, let me clarify one thing. My goal is not to ban buzzwords.

I just want to make sure we stop letting them run the show.

When the stage lights are on, you can see what’s happening.

You can tell who did what, what worked, and what didn’t.
You can finally see what good looks like in a way that different managers can actually agree on.

In practice, I believe that it all boils down to creating a small set of behavioral descriptors that leaders and managers can remember, while their people can actively use and refer to.

This turns important stuff like performance management, promotions, and succession planning into something useful and development-oriented rather than a bureaucratic, meaningless, session of Scrabble.

The camera test

It all starts with the simplest filter: could a camera capture it?

A camera can capture:

Summarizes the decision, owner, and next step at the end of the meeting.

A camera struggles with:

Shows maturity.

It’s much easier to give impactful and less personal feedback when behaviors can be truly observed.

It’s easier because instead of debating someone’s character, you talk about what happened.

If you catch yourself writing a trait, translate it into what the trait looks like in action.

Mature can become:

• Stays calm under pressure and keeps tone respectful.

• Owns mistakes quickly and proposes a fix.

• Raises issues early, without blame.

• Separates the person from the problem in conflict.

The bonus here is consistency.

Two managers can disagree about whether someone is mature. What does that even mean, after all?

On the other hand, both of these stubborn managers can usually agree on whether the person owned a mistake quickly and proposed a fix.

[If they disagree… oh, boy. I think you might have a completely different problem at hand.]

If you want to make this even more usable, tie each behavior to a moment managers recognize.

Things like:

• Meetings

• Deadlines

• Conflict

• Hand-offs

• Stakeholder updates.

Those are the scenes where work either moves or gets stuck.

Subscribe now

You need behaviors with sharp [clear] edges

A lot of descriptors fail because they are too broad:

Communicates effectively.
Drives results.
Builds relationships.

These phrases sound right and guide nobody.

What does effectively mean, in practice?
How many ways to drive for results can a person come up with? Whipping subordinates is an option, I guess?
What relationships? To what end? Why are we checking if people are having relationships in the company? What does the policy say about this? Oh, Lord…

A useful behavior has edges.

It shows what good looks like and hints at where it stops being good.

Edges reduce interpretation, and interpretation is where must things go off script.

For a project leader, drives results might look like:

• Defines deliverables, milestones, and owners early.

• Tracks progress visibly and follows up quickly on slippage.

• Escalates blockers early with options.

• Adjusts scope based on constraints and confirms changes.

Do you see what this does?

It makes results about the mechanics of delivery.
It also makes the behavior coachable: if someone is struggling, you can point to the exact step that person is missing.

Not bad, I’d say.

Edges also prevent weaponization.
[Sure, a blade has edges and can be a weapon, but let’s not get stuck on details, alright?]
Vague language can justify almost any decision after the fact. Clear behaviors make that harder because they force you to reference evidence.

[Quick self-check: if a behavior could apply to literally every job on earth, it probably needs sharpening]

Explicit levels are what you need

One set of behaviors for every level is a recipe for confusion.

The verb might stay the same, but the scale and complexity change.

Let me give you a practical example.
Communicates clearly at different levels might look like:

• Individual contributor: writes clear updates, asks good questions, confirms next steps.

• Manager: aligns the team on priorities, sets expectations, gives timely feedback.

• Senior leader: frames direction, makes trade-offs explicit, lands decisions under uncertainty.

This matters for two reasons.

First, development.

People stop guessing what that mysterious next level means and start seeing it in concrete terms.

That makes growth conversations easier and less emotional.

Second, succession planning.

Readiness becomes visible through scope, complexity, and impact, rather than confidence, volume, or charisma.

You can still value presence. You just stop confusing presence with capability.

When you define levels, avoid turning it into a novel. In my experience, a simple progression works well:

1. Self

2. Team

3. Cross-team

4. Enterprise

Most roles can be mapped to that without pain.

But do you know what’s painful? Rolling out this type of things without testing them first.

Ready, steady… TEST

I’ve made this mistake countless times.

After spending days and days figuring ouut the best behavioral descriptions, mapping out organizations, interviewing people, drafting, discarding, redrafting, it’s easy to end up with something you’re certain is perfect.

But that, unfortunately, proves nothing.

The only proof of quuality is whether managers interpret it the same way without a facilitator standing in the room.

So, test it.

Give the behavior list to managers across functions and ask for:

• One example of the behavior done well

• One example of it done poorly

• One sentence they would use to coach it

Then compare responses.

If the examples are all over the place, your wording still has fog in it. Tighten it.

Replace abstract terms with observable actions, and add context if needed.

This test also surfaces cultural and functional differences.

A behavior like speaks up can work very differently across your China and Italy-based teams.

You can keep the intent and clarify the form:

• Raises concerns early using facts and impact.

• Offers an alternative option, not just criticism.

• Disagrees respectfully and directly.

Now speaks up stops being a personality preference and becomes a repeatable practice.

Another useful move here is to collect a small library of examples once you finalize the behaviors.

Don’t imagine a giant encyclopedia.

Just a handful of good and not quite examples per behavior.

In my experience, managers love to copy examples - when they are relevant and well written - because they simplify their lives and make processes faster and coherent. A win-win-win.

Subscribe now

Make them real

Behaviors become real when they show up in decisions.

Employees learn the true standards by watching:

• Who gets promoted

• Who gets rewarded

• Who gets protected

• Who gets labeled “high potential”

If the company says collaboration matters but rewards the lone wolf who hits numbers while burning bridges, your descriptors become comedy material.

So weave behaviors into the moments that matter.

In performance check-ins:

• Which behaviors did you demonstrate this quarter? What evidence supports that?

• Which behavior was weakest? What will you do differently next week?

In promotion and succession discussions:

• Show me examples of influence without authority.

• Show me how they handled a tough peer conflict.

• Show me how they developed someone else.

This nudges the organization from storytelling to proof.

OK, now that they are real - you shouuuld make them easy

If you want the process to stay lightweight, make evidence easy to collect.

Encourage short, routine habits: meeting notes with decisions and owners, brief stakeholder updates, simple project milestones.

These create a paper trail of behavior without turning people into professional diarists.

And make sure you watch out for mismatches

Watch for the mismatch between what you say you value and what you reward.

That gap creates more cynicism than any single bad manager.

When you close the gap, trust rises - fast.

Share

Human, after all

When I do this kind of work with my clients, one fear always shows up:

Will this turn work into a checklist?

It can, if you write behaviors like compliance rules and use them like a stick.

If you use them well, behaviors can do wonders.
I’ve seen countless teams and businesses where they helped reduuce personal attacks and frictions, by removing the guuesswork and helping people formuulate and uuunderstand clearer expectations.

Writing is not enough, though.

You still [and forever will] need:

• Human judgement

• Context

• Managers who can think

[This last one might sound obvious but believe me, it’s not]

The behaviors are a shared language, not a substitute for leadership.

A practical way to keep it human is to treat behaviors as prompts for conversation, not as a scoring sheet for personality.

Ask managers to reference two or three behaviors that mattered most in a situation and describe what they observed.

That keeps the focus on what drove impact, not on ticking boxes.

Great lighting!

The fog machine will always be tempting.

It lets people sound smart without committing to meaning. It also quietly breaks fairness and consistency.

Stage lights are less glamorous, but they do the job.

Behavioral descriptors make work visible.

When they are designed well and tested properly, they can really tuurn arouund the way feedback is given, so that development is real and talent-related decisions are reasonable, visble, fair - and easier to defend.

And the good news is: You can keep the big words.

Companies love big words. Just don’t let them run the show.

Translate them into observable actions and test the language with real managers.
Then scale by level, and wire them into the decisions people care about.

That’s how you keep the performance on track, even when the cast changes.

Track one is the official story: values, competencies, promotion criteria, performance standards, succession plans. It’s tidy, and perfect for any PowerPoint deck.

Track two is how decisions actually happen: who gets trusted, who gets forgiven, who gets another chance, who gets labeled not quite there after that one slightly awkward meeting.

When these two tracks line up, people feel the place is fair.

When they don’t, the inspirational posters on the walls stop making sense.
You find managers who start guessing, and talent processes that become suspiciously dependent on who is speaking the loudest in the room.

This is why I believe we should all learn how to describe behaviours, really.

Let me explain.

When we describe behaviors, we use what in organizational development lingo we call behavioral descriptors.

Stay with me, I promise, this is the only truly boring part of this reflection. I think.

Behavioral descriptors are plain, observable statements of what people do, say, and deliver.

They take fuzzy corporate words and turn them into actions that anyone can spot.

They give employees a clear target, and managers something concrete to coach.

They help creating a culture the remains, even when the leadership team changes, because they provide structural support to systems that people need to use in the day-by-day.
[Systems like performance management, promotions, succession planning - just to name the ones impacting everyone in any organization from 3 to 50000+ employees]

I’m making an empirical claim from the field here.
In organizations where expectations stay abstract, I always see a variation of the following domino effect:

People fill the gaps with personal preference > Personal preference then gets treated like a standard > Fairness becomes a mood rather than a process.

So here is how you reverse the flow, if you want talent systems that are fair, consistent, and sustainable.

First of all, you need shared standards.
Shared standards require shared meaning.
Shared meaning comes from behavior, described clearly and simply enough that people can remember it.

Also, it reduces the number of meetings where adults argue about what synergy means with the seriousness of a constitutional convention.

Subscribe now

Who turned on the fog machine in the meeting room?

Corporate jargon works like a fog machine.

It adds drama and hides the stage.
Even better, it makes everyone look busier than they are.

Take, for example, the word ownership.

Everyone loves it.
But I dare you - ask around. You’ll soon find out that nobody wants to admit they have no idea what it looks like in daily work.

So people fill in the blanks:

• One manager thinks ownership means never escalating problems.

• Another thinks it means staying late until it is perfect.

• Another thinks it means speaking up confidently in meetings.

• An employee thinks it means taking initiative and hoping it does not backfire.

And through the fog on stage, enter the villain: The performance review.

Someone gets told they lack ownership.
They’re confused because they worked evenings for weeks.
The manager is confused because the person never flagged risks early.

Both walk away annoyed. Nothing improves.

That is the cost of fog.

A behavioral descriptor clears the air by giving the word a shape.

With a behavioral description, “ownership” becomes a short list of actions, such as:

• Flags risks early, with options.

• Delivers what they commit to, or renegotiates early.

• Keeps stakeholders updated when plans change.

• Closes loops by confirming decisions and next steps.

Now you can finally recognize it, and you can even coach it.

Different managers can apply it with less personal interpretation.

The words that break the system

When language is vague, three predictable things happen.

First, managers rate people based on their own definitions.
Et voila! You end up with an organization that thinks it has one standard - but it actually has 300.

Second, employees start optimizing for the preferences of their immediate manager.
They stop building transferable strengths and start playing local politics. That is a rational response to unclear expectations.

Third, calibration meetings become… well, just exhausting.
They become a tiresome sharade where leaders end up having to debate meaning instead of evidence.
Half the meeting disappears into the black hole of “What do you mean by [accountability]??”

It is too easy to think that this is an HR problem.
The sad reality, is that this is a reliability issue - that goes faaaar deeper than HR.

Performance management, promotions, and succession planning all rely on judgments.

Judgments become more consistent when the criteria are concrete and tied to observable behavior.

That is why industrial/organizational psychology created structured evaluation methods - and that’s why we keep insisting that we should use them.
[We are not just categorization freaks, I promise]

When you reduce ambiguity, you reduce variation in interpretation.
You also reduce the chance that confidence, style, or other bias like recency and halo become the deciding factor.

The point is that clarity helps us reduce the space where this bias can hide.

Labels don’t develop people

One reason jargon survives is because it sounds like feedback without requiring any actual feedback.

You need to be more strategic.
We want to see more leadership.
Your executive presence needs work.

Souunds good, eh?

The problem with these phrases is that they are quite convenient.

The even bigger problem is that they also leave the employee with nothing to do differently tomorrow.

People improve when feedback points to actions and outcomes.

Just look at this:

You need more executive presence

versus

In the meeting, you answered questions with long background, then never landed a recommendation. Next time, lead with your recommendation in one sentence, then add context if asked.

This is the difference between a meaningless label, and the beginning of a coaching conversation.

One is a label. One is coaching.

The first, vague feedback, easily lands as a personal judgement. Not nice, and fairly useless.

The second, way more specific description of observed behaviors is more likely to land as a skill discussion.

And let’s be real.

Skill discussions lead to learning.
Personal judgments lead to defensiveness, disengagement, or both.

So we’ve taken a look at vague, useless if not harmful feedback.

Time to move on to the next real villain on our foggy stage: Values.
[Of the company kind]

Make values usable

Values are where jargon goes to retire.

COLLABORATION - INNOVATION - CUSTOMER FOCUS - INTEGRITY

They’re good words.
Also completely harmless, until you try to translate them into daily actions.

A value only becomes useful when people know what they should do with it.

In practice, a company value should give a direction to people when their boss is not around, and they still need to make a decision.

Great values should always, always answer a very simple question: what should I do differently at work because of this?

Take collaboration.
In one company, collaboration means including everyone.
In another, it means moving fast and keeping people informed.
In a third, it means avoiding conflict.

Those three versions create three different cultures, even if the word on the wall is identical.

So you pick the version you want, then write it in verbs:

• Shares context early, before decisions are locked.

• Asks for input from teams who will feel the impact.

• Responds to requests within agreed timelines.

• Handles disagreements using facts and impact.

Try it yourself. I am sure you can add at least ten more descriptions like these ones.

But you pick the one the best embodies what your organization (your business, your team, your department, whatever) stands for, and suddenly, your leaders can coach based on it without sounding like a motivational poster.

Again, think about the difference between

You need to improve on your collaboration

and

I need you to share context earlier. When you wait until it’s final, people feel cornered. What if next time you send a draft update and ask for reactions.

Values move from wall art to real behavior when they become easy to describe and easy to spot.

So, let’s see how do you create great behavioral descriptions, shall we?

Write behaviors that people can remember

Even good behavior definitions fail if nobody remembers them.

I often see two extreme opposites - both common mistakes:

1. Writing them like a committee that got paid per word [Too short]

2. Writing them as long lists that no human will ever see [Too long]

Golden rule:

If you want adoption, you need recall.

That means:

[SHORT VERBS] + [CONCRETE OBJECTS] + [A BIT OF CONTEXT]

A few random examples:

• Flags risks within 24 hours when assumptions change.

• Summarizes decisions, owners, and next steps before meetings end.

• Shares progress updates weekly when timelines are tight.

These sounds like real work because they describe real work.
That is the whole point.

And please, keep the list short.
If every value has twelve behaviors, nobody can focus.

Pick the few that truly define success in your context.

Just like that old templar knight

So here we are.

Imagine yourself in a chamber. A Grail Chamber, to be exact.

All around you, cups and calices of all sorts - jewel-encrusted, gold, platinum ones too.

All beautiful, all very attractive.

And just as you are about to pick one, with sparkly stones that must be diamonds, you notice a sentence engraved on the side.

It says..

You should show more leadership.

By now, dear Professor Jones, you should know better.

Just drop it.

Behind it, a simple carpenter’s cup looks at you. And as you hold it, you hear the voice of that old, gentle knight whispering…

You have chosen wisely… the simpler cup, within the time we agreed upon, notwithstanding the many other luring but deadly options.

[OK I’ll admit that behavioral feedback is hard to make sound… epic, but I promise it’s effective in the long run]

This is it for Part 1 of this reflection on behaviors. Thank you for reading!

If this all feels slightly annoying, good.

Annoyance usually means you’ve started noticing the fog on the scene.

In the next part, we’ll get into the mechanics: how to write behaviors with clear edges, how to make them work across levels, and how to test them so managers interpret them the same way.

Stay tuned for Part 2:

Switch on those stage lights - How to write behaviors that survive real work

It’s a mix of amusement and a slight sensation of uneasiness.

Every week brings a new headline about a major firm quietly trimming staff and reshuffling business units.

Every other week, those same consulting giants announce some bold AI transformation program that reads like it was written at 2.00 AM by someone who already accepted their fate.

At the same time, AI tools keep getting smarter and easier to use. And cheaper, of course.

Even my most traditional clients are experimenting with AI (with very mixed results, if I may be honest with you).

They might not trust it fully, but they are definitely touching it… with great interest.

Consulting has always been a strange hybrid of intellect, theatre and labor.

The intellect part is safe.
The theatre part is eternal.

The labor part is what AI now threatens.

Let’s be honest: a lot of this labor has been the core of the consulting business model.

Traditionally, consulting firms spend tremendous amount of (billable) hours gathering data, cleaning spreadsheets, coding interviews, summarizing documents, producing analysis that looks sophisticated but is really advanced pattern recognition with a suit.

Now, the problem is that AI does not get bored doing these tasks - and doesn’t charge for them.

That alone should make anyone in our field stop and think.

I see two reactions among fellow consultants.

The first group is convinced AI will never replace human judgment.
They say this with the same confidence taxi drivers had in 2011.

The second group is quietly updating their LinkedIn headline to Advisor.
It does sound a bit more future-proof and I’ll admit, I am probably part of this group.

Both reactions have some truth in them, but I feel like we may be still missing the point.

I do believe that AI will replace most consultants.

[Note: My knees are slightly shaking as I review this draft before posting. But I’ll embrace the fear and keep this statement as it is]

The question is how consultants adapt when more and more clients build internal capabilities that used to be our bread and butter.

That brings me to my own reflection on what comes next.
Or what might come next, as these days you may say something that sounds obvious today, and tomorrow you’ll sound like a fool.

Subscribe now

Your spreadsheet has become sentient. What now?

The rise of accessible AI tools is shifting the balance between clients and consultants.

Clients now run their own diagnostics.
Some of my clients process their own HR data at a speed and depth that, five years ago, required a full consulting team.

This is already happening, it’s not hypothetical.

As companies adopt these tools, the entry point for consulting changes.

We will be called when the client already has a report, a dashboard, or a summary.
[Goodbye, nights spent fixing PowerPoint animations?]

This means that we, the consultants, will be needed for our judgement.

Clients will need us to help them see what the data does not show.

They will rely on us to translate insight into action, culture, behavior, alignment and decisions. They will look for partners, not vendors.

This creates a brutal consequence for the industry.

Consulting firms that relied on labor volume will struggle. Some already are.
The ones who built their identity on frameworks and templates will fight to stay relevant.

The ones who differentiate through expertise, clear thinking and humility will thrive.

The market is transitioning from do the work for me to help me make sense of the work I already have.

To survive, consultants need to become true experts.
[It was about time, if you asked me]

Consultants will need to be much more than PowerPoint experts.

They will need to truly study patterns across industries, all the while staying ahead of the academic and technological curve.

Consultants will need really know how to read organizations in motion.

The model will need to transform: From labor to wisdom.

Which sounds romantic until you realize it involves hard work and years of learning that nobody bills for.

Congratulations! you’re now and advisor. Please know everything.

Becoming an advisor means the bar for credibility goes waaaay up.

If a client’s AI can generate a decent competitor analysis in five minutes, then your value is no longer speed or detail.

Your value is interpretation, powered by courage.

The courage to say the uncomfortable thing and recommend a different strategic direction, even if it means challenging a leadership team that already thinks it knows the answer [because AI told them so].

The shift from consultant to advisor might free us from the lower-level tasks, but it also exposes us.

There is nowhere to hide behind Excel. When clients can generate their own baseline, a good consultant must be able to ask themselves:

What do you see that they don’t? How do you help them move?

In practical terms, consultants must do three things.

First, narrow their domains.
Nobody trusts a generalist advisor.

Second, build a point of view.
Not a rigid ideology, but a coherent lens.

Third, get better at the human side.
AI cannot coach an executive through the discomfort of behavioural change, nor can it navigate politics.
In my field, for example, AI cannot replace the subtle social reading that defines real organizational development.

This future rewards consultants who bring insight and can navigate nuance.

The ones who obsess over tools and deliverables will find themselves negotiating with algorithms that work faster and cost nothing. Not a nice place to be.

What I’ll keep watching

So, here are a few things that I will keep an eye on - and if I may be as bold as to make a suggestion to you, fellow consultant reading my ramblings: You should keep an eye on this stuff, too.

1. How fast companies truly adopt AI at scale
   Many leaders talk about AI, but only a few redesign processes and roles to unlock real value.
   I will watch who is actually restructuring workflows, not who is making announcements.
   Some of my clients are actively looking at re-designing their systems, and I reckon this trend will only increase in the near- and mid-term.

2. How consulting pricing models respond when labor decreases
   The industry will need to shift from selling effort to selling clarity, which demands a new kind of pricing courage.
   I have started experimenting with value-based fees and may try to explore advisory retainers, with mixed results.
   But I am working on this before the market forces my hand.

3. Whether advisors become part of leadership ecosystems rather than external providers
   Complexity is rising, and I have the strong sense that as this happens, trusted thinkers may end up sitting closer to the decision table than ever before.
   Build long-term relationships and demonstrate judgment in moments of ambiguity will probably become increasingly important to help leaders see consultants as partners rather than vendors.

4. How expertise is validated when everyone suddenly has AI enhanced insights
   Signal and noise will collide, and credibility will depend on pattern recognition, not volume of output.
   I would encourage any consultant to double down on developing a clear point of view supported by real-life cases and comparative experience, and make it visible in their work.

5. How the human side of organizations evolves as automation increases
   Work will get faster, but motivation and meaning will still depend on relationships, culture and leadership.
   I bet that a competitive edge will be honing coaching, facilitation and sense-making skills because the premium will migrate toward the human tensions that AI cannot resolve.

So these are the 5 things I am currently reflecting [read: obsessing] on.

I am probably missing some other absolutely critical aspects of this transformation so if I do, feel absolutely free to pint them ouot. I’d love to hear your take on this, and to learn what others are doing in this times that are definitely A-changing.

The consultants are not okay, but the consultants who adapt might be better off than ever.

It was a fantastic event, with several panels exploring the impact of AI on different aspects of business, organizations, and individuals.

Attendance showed how relevant the topic is today in China: Local and foreign professionals, cross-industry and cross-functional experts, in-house employees and consultants, all flocked to a busy venue on the river to listen and talk for an entire afternoon and into the evening.

These days, AI is a recurring topic in all my meetings with clients, roundtables, industry events, and even chats over coffee with fellow consultants and service providers.

When the conversation with my HR clients inevitably touches upon AI, the most common sentence I hear is

We need to embed AI into our processes so we can be more strategic.

And every time, I can’t help but notice how AI seems to be perceived as a sort of magic thing that will lift everyone out of admin, menial work and into a glorious future of strategic thinking and meaningful conversations.

There is only one small problem.

AI is not going to magically transform HR if it is confused, underpowered, and in order-taking mode as it, unfortunately (and with few exceptions), still is.

AI is a magnifying glass, that makes things bigger and faster. Including the bad things.

As long as HR is spending most of tis time firefighting, AI will help it fight slightly faster fires.

The point is, when it comes to HR, we should not rush to upskill on AI. We should ask a way more basic question:

Can HR even tell the difference between good work and noisy work, and make decisions with confidence?

Only then does AI become interesting.

Share

That shiny robot in the corner

Right now, many HR teams are in experiment mode.

Some of my clients are playing with chatbots, hoping that they may help better connection with the employees.

Others are buying AI tools for recruitment.

And while their internal communication decks - and their AI vendors’ flyers - use big, bombastic words, the reality looks sadly more like this:

The chatbot is rewriting job descriptions that were already unclear, making them even worse.

The AI screening tool that ranks candidates uses criteria no one has fully agreed on - and as a result, nobody is able to see if the shortlisting even makes sense.

Another one has now created a talent dashboard and has no idea why it looks the way it does. Good luck presenting it to the CEO next week.

I often see leaders treating AI like a new colleague who just joined and, by some miracle, already knows the job better than anyone else.

But AI has no judgment.

No context.

No clue about your office politics.

No knowledge whatsoever of your history of failed initiatives,

Is not aware of the fact that one of your top performers actually burns out their entire team every six months.

That part is still human work.

An old story with a new branding

Automation is not new for HR.

Payroll, leave tracking, basic reporting, applicant tracking systems. Many of these have been automated for years.

AI is simply turning up the volume on this stuff.
It’s making them faster and more scalable. In the best cases, it’s helping reducing friction.

If your process already makes sense, this is good news.

If your performance framework is clear, your competency model is solid, and your interview guides are thoughtful, then automating parts of them can save time and reduce human error.

If the above is true, as a consultant I would have no hesitation in telling you: OK, time for some AI upskilling.

But the problem is when you’re faced with the opposite.

If your process is vague and misaligned, or built on the notorious we have always done it this way, AI gives you faster confusion.

This is why, when I hear that a client wants to automate first and clean it up later, I tend to freak out.

Because later rarely comes.

Acceleration: When HR finally gets to think

Now the interesting part.

AI can become the world’s most enthusiastic HR intern.

It can draft policies, summarize surveys, pull themes out of messy text, prepare first drafts of communication, and crunch data in ways that used to require a friendly analyst and a lot of coffee.

Used well, this could change the daily work of HR.

Instead of spending three days preparing slides, HR can spend those three days in conversation with stakeholders, exploring trade-offs, asking better questions, and pushing for decisions.

Instead of manually reading 200 comments from an engagement survey, HR can ask AI to cluster themes, then - and I will never stress this enough - use human judgment to interpret them, challenge them, and connect them to the real politics of the organisation.

When I say acceleration, I mean that HR should think of AI as a companion that helps getting back time for meaningful activities.

But that only happens when HR already knows how to think - and act - strategically.

The robot intern needs supervision

AI needs supervision.

It needs supervision when you’re setting it up (if it was a robot intern, you’d say onboarding), but also, and most importantly, throughout its tenure.

It means, continuous supervision.

in practice, this means, that when HR is using any AI tool, someone will need to:

• Make critical decisions about which data the tool is allowed to see

• Question the criteria used for scoring and evaluation

• Check output for bias, fabricated information and conclusions, nonsense, and/or unintended side effects

• Review and translate AI recommendations into stuff that makes sense for the real humans, in a real culture, that will have to live and deal with them

Because if HR is weak, anything produced with AI becomes really loud. And if it’s wrong, or if its negative effects were not anticipated and corrected, well, that’s an obvious problem.

A strong HR team treats AI like a smart intern: helpful, fast, sometimes brilliant, sometimes very wrong.

An intern can prepare the first cut of a talent review grid.
Someone senior still needs to look at it and realize that the output doesn’t match what they see on the ground, and that more and deeper digging is needed.

Just like a smart, enthusiastic intern, the value of AI’s contribution depends on the strength of the real, human filter around it.

First, fix HR. Then, plug in the tools

So what does “strong HR” mean in this context?

Spoiler alert: It has little, if nothing to do with AI.

In my experience, strong HR looks like this:

• Understands the business model, not only HR processes.

• Can hold its ground with senior leaders, ask hard questions, and say no when needed.

• Knows what good performance looks like, beyond the rating scale.

• Has a clear point of view on talent, potential, fairness, and growth.

These four points do not apply only to the top HR leader in your organisation. in a strong HR function, every member of the HR team should demonstrate these behaviors, pitched at their level.

When these foundations are weak, AI projects end up like nice-to-have-but-a-waste-of-money sort of experiments that only erode HR’s credibility and impact.

When they are strong, AI becomes a powerful helper.

In my client work across industries, the most successful “AI in HR” stories have a similar pattern.

Instead of starting by looking for a tool, the company started with an HR team that already had a voice at the table - and knew when and how to use it, and for what purpose.

They already had an HR team capable of conducting analysis, have conversations, and setting up strong learning cycles.

With AI, they are making the analysis even sharper, the conversations even more meaningful, and the learning cycles faster.

With AI, they are scaling what they were already good at.

So what do you do tomorrow?

If you are an HR leader or a senior executive reading this and nodding along somewhere between curiosity and slight panic, here are some practical starting points you can act on right away.

I do not mean to imply these are ultimate solutions, and you must make sure you adjust these ideas based on your reality, of course.

1. Run a brutally honest HR skills check

Gather your HR team and ask a few simple questions:

• Where do we already add clear strategic value?

• Where are we still acting as admin support with better English?

• Which decisions are we afraid to own?

Write the answers down. And prepare a plan to get to a point where the answers are clear, solid, and true.

Once you’re there, go out and look for the most approrpiate AI tool to make them even better.

2. Pick one problem, not one tool

Instead of saying: “We should use AI in recruitment,” try “We waste too much time screening CVs for role X.”

Work from the problem backward.

Then see where AI might accelerate a step that HR already understands.

3. Design the supervision, not only the workflow

For every AI use case, define:

• Who reviews the outputs.

• How often they challenge the logic.

• What happens when AI and human judgment do not agree.

Put this in writing.

Treat it as seriously as you treat approval rights for promotions or terminations.

4. Invest in HR thinking skills

Before sending everyone to “Prompt Engineering 101,” build confidence in core consulting skills:

• asking better questions

• framing problems

• connecting people issues to business outcomes

• reading data with a skeptical eye

A confident HR professional with a basic AI toolkit is far more dangerous (in a good way) than a prompt wizard who has no view of the bigger picture.

5. Talk openly about ethics and risk

Bring up the uncomfortable questions early.

• How will you avoid hidden bias in AI-assisted hiring?

• What level of transparency will you offer employees?

• Who signs off when AI is involved in decisions that change a person’s career?

Do this before someone in Legal walks into your office with that look.

AI will keep growing in power and presence, and vendors will keep showing fantastic demos.

Boards will keep asking why HR is not doing more with AI, as all the industry and big consulting reports keep on saying.

The teams that navigate this well will be the ones with the courage to treat AI as a very fast servant, not a silent boss.

To do this, build strategic skills. Empower, truly, for better outcomes, not just nicer processes.

Strengthen HR first. Then let AI run.

Not work-late-once-in-a-while fast, but structurally fast. Built-for-speed fast.

While Western firms often talk about scaling, their Chinese counterparts talk and act in terms of speed.

While many Western organizations still chase scale through capital and control, their Chinese counterparts chase speed through iteration and integration.

Especially over the past couple of decades, the Chinese business mindset has evolved into a system that is in a sense, wired for conversion and striving for movement and participating in the game, rather than focusing on perfection.

Bringing together a dozen conversations I had with local and foreign executives over the past few weeks, today I want to try and unpack how China’s business engine manages to go from spark to scale in less time than it takes a Western company to finish its second steering committee meeting.

Subscribe now

Ideas go in, stuff goes out - at record speed

I’ll start with the math.

According to the Global Innovation Index, in 2025 China ranks 19th for innovation inputs (R&D spend, human capital, infrastructure) but 5th for outputs (patents, products, designs).

In other words, it’s quite clear that when it comes to innovation, China is playing the output game better than most, making sure that what matters is not really how much goes in, but rather, how quickly something useful comes out.

The system rewards throughput over preparation.

It would be easy to dismiss this as sloppy.
But that would be a mistake, because the way many executives describe it, it sounds more like it’s simply optimized for velocity.

[It might be worth adding here, that China’s ranking on GII was 23rd for input and 7th for outputs in 2024 - gaining 4 and 2 positions respectively in one year]

Clusters, not castles

Come to China and you’ll soon figure out there’s a real passion for industry clusters.

These clusters are, to put it simply, geographical concentrations of companies in related fields that can leverage on:

1. Specialization within a specific industry,

2. A comprehensive ecosystem of supporting industries and infrastructure,

3. Consistent and deliberate government support (with dedicated policies and special economic zones designed specifically to help the clusters’ development), and

4. A push towards integration across city and provincial boundaries to generate a more efficient economic environment.

What this boils down to, is that these clusters offer a tremendous advantage: A science and technology cluster for example, integrates an obviously dense population of engineers with suppliers, factories, logistics, and customers.

This means that supply chains are so well embedded that things can move really fast.

As a friend of mine put it - less emails, more human interaction right where and when you need it.

And this, in turn, highlights another aspect of China’s speed: Proximity.

When the factory is next door

The integration of supply chains has another visible result: R&D and manufacturing become much closer.

This means that the gap between an idea (say, a new battery chemistry) can be tested, tooled, and scaled within weeks because the plant is essentially “across the street”.

The design → tool → pilot → ramp cycle that can take traditional Western firms quarters often takes Chinese ones months, if not weeks.

In many ways, this integration dissolves the old corporate sequence of research first, production later.

I often realize how, in China, research and production are not two subsequent steps but rather elements of the same sprint.

Geography turns into strategy.

And then, comes IP

[An important note here: IP protection in China is a very complex legal and policy space on which I can’t comment due to my limited knowledge on the subject - but I would welcome any expert opinion to integrate this piece!]

As I mentioned earlier, China seems to be able to leverage on a very positive focus on innovation outputs - and patents reflect this.

I was recently told about the concept of utility models - as I am not an expert on this, I will over-simplify by describing them as a sort of fast-track intellectual property rights that protect useful tweaks and not just groundbreaking discoveries.

Regardless of its intricacies, this preference is quite telling because it seems to indicate that Chinese firms are comfortable with early patenting, frequent iterations, and later improvements.

It’s a sort of agile IP - a culture of small bets that move faster rather than big bets that might move… never.

A highway for customer feedback

As anyone in business knows, the end user’s feedback is after all an important (if not the most important) aspect of any strategy.

And the Chinese digital environment reinforcement of this concept has evolved to a level that remains hard to grasp if you’re not immersed into it.

China’s digital services platforms, ranging from payments to delivery, from services booking to restaurant and product reviews - powered by super-apps like WeChat, AliPay and Meituan to name the most famous ones - means that experimentation happens in real time.

But what does this mean, in practice?

it means that a product can hit the market, get feedback, and be tweaked at record speed.

When infrastructure itself can move at such a speed (logistics, payments, supply chains), companies can afford to test, fail, and fix at a speed that for many European or American counterparts is, simply, impossible.

Leave a comment

It’s chaotic, but it’s OK

Let’s be real. China’ business scene can feel very chaotic. But, I argue, it’s functional chaos.

A system that rewards speed has to tolerate messiness in its many forms. As an organizational development observer, I often realize how Chinese clients deal quite comfortably with overlapping roles and concurrent projects, even unfinished processes.

I will admit, I still prefer having a level of clarity (it must be my European upbringing!) and to be even more frank, I know that many Chinese employees would also like a little more stability and less ambiguity in their organizations.

But to stick with this trajectory of West vs East I am finding myself more and more trapped in as I ramble through this reflection: It’s still scale versus speed.

A good friend of mine, a Chinese senior leader working at one of the biggest European FMCG firms, describes it like this:

Western enterprises, built on scale, tend to seek order: stable processes and risk controls.
The Chinese system, built on speed, accepts disorder as the price of movement.

In my Italian brain, one looks like a cathedral. The other, a bazaar.

Both have logic.

But when change accelerates, I have the distinct feeling that the more flexible bazaar may sometimes beat the beautiful cathedral.

So, what.

What can foreign firms learn without trying to become more Chinese?

For starters: Speed does not equal recklessness.

Speed means designing structures that enable faster feedback loops.

And this can look very different from one company to the other.
Some of my clients are co-locating teams, other prefer institutionalize knowledge and information sharing, I’ve helped firms with shortening decision-making chains, and I’ve seen many trying to really break down silos across functions.

But this also means, rethinking risk.

Today, the bigger risk lies in inertia. Failure should be seen as a necessary aspect of innovation and speed.

Many Chinese firms, and especially those who are now extending themselves beyond the domestic market, are consistently making small, (somewhat) reversible bets, and they learn fast by compounding the lessons they take from them.

And now, to turn everything around

Scale and speed are not enemies.

The way I see it, they are just different phases of the same game.

Fast iteration eventually builds scale. Look at BYD.
The Chinese EV manufacturer is now expanding its global footprint, and this didn’t emerge from chaos alone, but from hundreds, even thousands of micro-decisions that snowballed.

There are also many western multinationals who have learnt to inject speed moments into their scale system, and thanks to faster decision-making cycles and more local autonomy, are seeing remarkable payoffs.

I guess that the real art is to know when to sprint and when to slow down and stabilize.

You may be thinking, at this point, that I am just another China enthusiast. I would like to take half a second to rectify this possible misunderstanding. I am not.

On the topic of knowing when to sprint and when to stabilize, for example, I think that Chinese emerging companies may also have to learn something from western counterparts - the difficulties companies like BYD and NIO are facing in their global expansion should be sufficient proof that sometimes you may be sprinting a bit too far.

But with the pace of change we are all experiencing, I truly believe that the ability to accelerate learning, not just production, may well become the ultimate competitive edge.

The hum of motion

As I meet people, discuss with clients, and explore the business environment in Shanghai, I find it very easy to feel the hum of the collective rhythm of experimentation, iteration, failure and learning that rarely stops.

But I also see that the system is far from perfect.

Speed produces casualties.

These days obviously the thought revolves around the seeming conundrum of the Chinese government’s bet on innovative technologies as a new focus of the planned economy.

The context is not the easiest.
Youth unemployment, talks of involution, and fear for job loss with the advent of AI-enhanced automation impacting knowledge workers are but a few examples of the many challenges that government, businesses, and individuals are faced with.

But there seems to be a general agreement on the underlying principle: Motion creates opportunity. Sluggishness* creates stagnation.

*[Note from a non-native speaker: is this actually a word?]

Foreign firms in China often marvel at this energy, sometimes fear it, occasionally misunderstand it.

But they can also learn from it.

Because while the West perfected scale, China perfected movement.

And I have the strong feeling that the winners in the end will be those organizations that found a way to integrate both sides of this spectrum.

And when it comes to foreign firms, taking a better look at Chinese companies and realize that they may be fast, but not necessarily furious, is going to be the only way they can try and remain competitive… in the long run*.

*[Note: Pun intended]

[Note: Unfortunately, as you can imagine, I never got paid for that and as a result, I am still writing from my apartment in Shanghai]

As a long-time organizational development consultant in China, I have attended more HR meetings than I can count.

I’ve seen slides quoting Confucius and heard promises of enhanced performance and inclusion. I’ve sighed over dozens of statements of strategic partnerships with the business. And on, and on, and on the jargon-filled list goes.

But after all these years, one question remains unresolved:

Why does HR, despite its ambition and hard work, still struggle to gain full trust from the business line?

This reflection is my attempt to diagnose the persistent (and at times, quite tragic) ailments of HR in China, and propose a few practical remedies.

Let it be very clear: these reflections are not meant to describe all HR teams in China. Many HR teams I know are highly competent and impactful, and globally competitive.

But I still see many others who are still stuck between old habits and new expectations, ending up doing menial admin tasks while the world around them changes, and technology marches into the future, leaving them behind (and, I fear, jobless soon).

So here I am, writing my own report, honoring Chinese bureaucratic rhetoric to the best of my abilities: Four Problems and Two Actions.

Subscribe now

Part I: Four Problems

One: Insufficient knowledge of the business

<< It is imperative to enhance HR’s comprehension of business models and market mechanisms to align human resources strategy with enterprise development goals. >>

To put it simply, HR may know the policy, but not the profit.

In numerous organizations in China, HR still functions as a parallel universe.

They are fluent in internal processes but detached from the operational heartbeat.

When the sales director talks about market share or the COO mentions gross margin, HR nods politely, then retreats to headcount reports and payroll.

This lack of business fluency undermines HR’s credibility.

HR still needs to learn how to truly talk the language of productivity, performance, and impact on the bottom line.
I am not saying that administrative tasks, cost-cutting, and cost-saving should be ignored. The problem is, they make little sense if they become sole goals of the function, rather than enablers of the business.

Some time ago a GM of a plant just outside Shanghai put it this way:

“When HR stops telling me what’s fair and starts telling me what’s profitable, then we’ll really talk.”

Point taken.

Two: Reactive needs analysis

<< Human Resources must strengthen forward-looking talent diagnosis and early warning capabilities to anticipate organizational challenges before they occur. >>

HR should stop waiting for problems to land in the inbox.

I love this quote from one of my clients:

“HR only runs when there’s smoke.”

In too many companies, HR operates as a firefighter rather than a detective.
Only when turnover spikes or morale dips does HR swing into action.

Proactive needs analysis, on the other hand, requires HR to step outside its comfort zone, listening deeply to what’s not being said and scanning for signals to anticipate crises.

This mindset shift implies an ability to look for answers to questions like “What’s going to break next quarter?” or “Where will our next leadership gap appear?”

I have to say, this is not an easy sell in a culture where HR is mostly focusing on cost-saving and compliance, and where no news is considered good news.

The most performing and innovative businesses I know have this element in common: An HR team that is not just sitting there passing paperwork around.

Three: Limited engagement with the business line

<< Human Resources shall strengthen field engagement and joint collaboration with frontline departments to ensure policies are grounded in operational reality. >>

HR should spend less time in the office and more time where the work happens.

I recently looked at a tragedy unfolding within an HR team as they could not agree on who within the team should be responsible for the company’s training programs.
In the meeting room next door, the head of sales for the Greater China region was discussing with the CEO the fact that a recent change in policies was about to slash their market share in the domestic market.

This is the gap I’m talking about.

Many HR teams still act like internal auditors.
They issue memos and collect reports.

But today, that’s no longer enough. HR needs to truly engage with the business.

This means, going to the production floor and see how things are done.
It means visiting sales teams and understand what they’re actually doing and what challenges they’re facing.

Partnership does not mean getting the proverbial seat at the table.

Surely, the HR manager who sits beside the sales VP during planning discussions has a very different influence from the one who waits for approval via email.

But the HR manager who comes to the same meeting with ideas on the plan itself, and proposals on what can HR do to support becomes a leader in their organization.

HR needs to stop being the “policy police” and start being the “solution partner”. Only by doing this, will HR build their credibility.

The most respected HR professionals I know are able to identify and solve business problems.

Four: Weak change management expertise

<< In response to ongoing transformation, HR must enhance its professional capability in leading, facilitating, and sustaining organizational change. >>

In other words, circulating an email about transformation is not transformation.

In one company I tried to work with some time ago, the change communication plan ended up being literally a poster in the elevator. Three months later, no one remembered what had changed. Not even HR.

[Note: I think the poster is still in that elevator]

China’s corporate landscape changes faster than trends on TikTok.

Mergers and restructures, digital transformations, all common stuff.

But in China, too many HR departments still treat change management as a communication checklist: send announcement, hold town hall, update posters.

Then, wait. And forget.

Real change management involves understanding resistance, mapping stakeholders, measuring impact, and sustaining commitment.

HR should be the custodian of these processes. And it ends up being a messenger (and the type of messenger others tend to shoot at - not cool).

As far as I know, too few HR professionals in China receive formal training in change management.

And this results in detachment, lack of credibility, and a further erosion of trust.

Subscribe now

Part II: The Two Actions

Action One: Train HR in proactive consultation

<< Comprehensive capacity-building programs shall be established to cultivate HR’s consulting mindset, strategic thinking, and facilitation skills. >>

HR must learn to ask better questions and give better advice.

The most effective HR professionals I’ve met think like consultants.

They can seek issues and diagnose them, listening to colleagues in other functions to validate their hypothesis, and they end up being able to problems clearly before proposing solutions.

This skillset is a tricky one in China, as the educational system does not really cover this crucial aspect of critical thinking, leaving individuals to learn this by themselves.

This means, these skills must be developed deliberately.

The biggest problem here is that traditionally, HR in China focuses on compliance, process, or systems.
Admin HR, to put it bluntly, is a very easy entry point for anyone - do you know how to use Excel, and can you learn how to fill out payrolls at the end of month? Welcome to your HR career!

The problem is that after that, very few companies invest in augmenting their HR function’s knowledge and skills.

And they end up with HR teams that will soon be replaced by mindless, but very efficient and cost-effective, robots.

The problem is, companies are already doing this - after all, automation has started taking away menial HR tasks long time ago, well before we called it AI.

The impact, however, is not on the HR function only - the impact is on businesses, because they are also losing a whole bunch of people that is (should be) positioned to do what no business leader can truly do: Connecting, engaging, coaching, exploring, and building workplaces where everyone can thrive.

So, companies should train HR on proactive, consultative skills.

By doing this, HR can stop being reactive and can step up to an initiator role. So it can stop begging for the seat at the table, and can gain it by contributing.

Action Two: Embed business Impact in HR Performance

<< Performance assessment mechanisms shall integrate HR’s contribution to business outcomes, emphasizing value creation over cost reduction. >>

Stop rewarding HR for cutting costs. Measure HR by how it helps the business grow, not just how it saves money.

For years, HR success in many organizations has been measured by efficiency: Things like reducing headcount, and cutting turnover costs.

The unfortunate result is HR departments that excel at saving money but struggle to justify their strategic worth.

I always ask HR leaders, how do they connect initiatives and HR activities they’re sponsoring with business results and unfortunately, that connection is not always there and when it is there, there’s a painful lack of monitoring and measuring systems.

It is time to expand the scorecard.

HR performance indicators should include productivity gains, capability growth, leadership pipeline strength, and employee experience improvements.

We need HR’s own performance evaluation to reflect its business impact.

We need conversations with finance to evolve from “How do we spend less?” to “How do we perform better?”

To do this, HR needs to step up their game in terms of skills, knowledge of the business, and strategic, critical thinking skills, so that HR activities can be judged by value creation, instead of meaningless (and often harmful) austerity.

A parting thought

I believe Chinese HR stands at a crossroads today.

The admin backbone is definitely there, and will soon be further enhanced by new developments in technology.

The potential remains enormous, even if I think there’s a long road ahead for HR to transform across the board.

This Four Problems and Two Actions framework may sound a bit silly, but the message i want to send is serious.

HR cannot earn respect through slogans or certificates. it needs to earn it through results and relevance.

If HR strengthens business understanding, anticipates needs, engages the frontline, and leads change (with training and performance systems that reinforce these priorities), then a new generation of HR professionals will help define its future.

And when that day comes, perhaps the next official report will be titled:

<< Zero Problems and Infinite Actions: The Glorious New Chapter of Chinese HR. >>

According to a survey conducted by his firm, consumers in first-tier cities are turning increasingly toward domestic brands, while those in lower-tier cities are showing growing enthusiasm for foreign brands.

Based on this, he seemed to point at a possible strategy for foreign brands in China: Double down on lower-tier markets, where interest seems stronger.

As I listened to him, I couldn’t help but wonder whether the conclusion isn’t a bit too short-term, and whether there may be broader and longer-term considerations to be drawn from the data.

[An important note before we continue: I am not a marketing expert, and my understanding of marketing and branding is foundational. This reflection is based on my understanding of these topics and does not presume to be a fully coherent, solid analysis.
In addition to the above, it’s only fair to clarify that I am in no way trying to debunk the presenter’s views, as I am zooming in on something he covered in 3 minutes as part of a much broader and well-researched presentation.
In fact, if you have better data, or a different perspective on what I am about to cover, please do reach out or share in the comments section! OK, having made the necessary disclaimers, we can now proceed]

What made me pause, is that there is a couple of other major trends in China which I can’t help but consider connected with, if not even contributing, to this shifting consumer behaviors.

The first trend has to do with a shift - of people, before loyalties.

When people move, loyalties travel

One demographic trend that has emerged over the past few years in China is that young, educated professionals seem to be leaving the big cities.

First-tier cities like Beijing, Shanghai, Shenzhen, Guangzhou are still aspirational, but they are also… well, exhausting.
The cost of living is high, competition is relentless, and the lifestyle seems to be increasingly viewed as unsustainable. Glamorous, sure, but unsustainable.

As a result, many of these professionals are relocating to lower-tier cities.
They may move for family reasons, work flexibility, or simply for a different pace of life. Smaller cities now offer more than decent infrastructure, growing industries, better environment than just a few years back, and so on.

So here’s my first question.

If this migration is indeed as significant as recent census data suggests, could it be that the “foreign brand lovers” are not disappearing, but simply relocating?
What looks like a shift in consumer taste might, in part, be a shift in population composition.

If that’s true, then the idea that lower-tier cities are “emerging markets” for foreign brands might be a little misguided, and misleading, because these are not entirely new markets. They’re extensions of first-tier consumer groups, settling somewhere else.

And this brings me to the second trend we have been observing for some time.

A maturing consumer base

Consumers in China’s first-tier cities have been exposed to foreign brands for decades.

They have experienced the early fascination with Western goods. Then, they went through the explosion of international luxury. And now, they’re active spectators to the rise of domestic competitors that are both high-quality and culturally confident.

They’ve gone through an entire learning curve.

We experienced a very similar curve in Italy, too.
My grandfather, just a kid during and after the Second World War, always used “American” as an adjective denoting products that are impressive, shiny, top-quality, and reliable (in his mind, at least).
Fast forward to the ‘90s, a young Fabrizio watching TV ads was hearing again and again that product X and Y, and Z were Made in Italy and understood that they were relatable, safe, and high quality. Regardless of the veracity of these statements, of course.

China seems to be making no exception to a very natural maturing process of the market and its citizens, the consumers.

Let’s go back to these younger Chinese consumers, then.

They have developed a strong sense of discernment and therefore, they evaluate foreign brands less for their foreignness and more for their relevance.
For over 30 years, they have grown up with an abundance of choice, so novelty is no longer enough.

Meanwhile, domestic brands have learned fast. They’re not copying anymore, as they’re now competing on design, quality, and emotional resonance because they have become better at speaking to cultural sensitivities.

The result is a consumer market that has matured beyond fascination with imported prestige.

So here is my second question.

If Chinese consumers are maturing, then the declining attraction of foreign brands in first-tier cities shouldn’t be considered just a natural evolution?

If this is true, this trend might simply point out that Chinese consumers who have been exposed to foreign brands long enough, and with local competition now entering the fray with a stronger positioning and better narratives, have outgrown the simplistic logic of foreign = better.

Subscribe now

The mirage of chasing growth

If foreign brands take the marketer’s advice and rush to invest in lower-tier cities, they may indeed see short-term results.

In fact, many foreign brands already are - luxury houses have been opening flagships all over the country in second tier cities, and many, many smaller firms have decided to focus on second- and third-tier urban centers, too.

But what happens five or ten years from now, when the consumers in those locations go through a similar maturity curve?
(To be honest, I wouldn’t be surprised if the timeline will reveal itself to be even shorter - it is China we’re talking about, after all)

Lower-tier markets will mature too, and when they do, their consumers may turn to domestic brands for the same reasons as those in first-tier cities: Proximity, cultural fit, a sense of ownership. Let’s also throw national pride in the mix.

So, the question for foreign brands might not be “Where should we go next?” but rather “How do we stay meaningful over time?”

Relevance is a moving target - for organizations, too

As I was writing this reflection, I realized that there might be an interesting connection with something I deal on a much more regular basis: Employer branding.

The fluidity we observe in Chinese consumers, shaped by lifestyle changes, social mobility, and identity shifts is, in fact, not limited to purchasing preferences.

From brand loyalty to employer loyalty

Companies often think of brand in terms of marketing, but the same principles apply internally.

Employer brands compete for talent in much the same way product brands compete for consumers.

Increasingly over the past years, they rely on relevance, trust, and emotional connection.

If first-tier consumers are maturing beyond foreign brands, we might ask: are first-tier professionals maturing beyond foreign employers too?

Multinationals once had a powerful draw in China.

They offered better pay, international exposure, and a sense of prestige.

But that advantage has definitely eroded over time.

Domestic companies now offer comparable (if not better) packages, faster career progression, and a growing sense of purpose tied to national confidence.

For many young Chinese professionals, working for a domestic company no longer feels like a compromise.

This in a way echoes the same dynamic we can see in consumer markets: a shift from aspiration to identification.

If foreign companies interpret this trend merely as a “talent drain” toward local firms, they risk repeating the same mistake marketers make when they chase lower-tier cities, because the issue is more about meaning and purpose rather than location, pay, or career (although, let’s be real, these three elements still play a crucial role).

The organizational mirror

From an organizational development perspective, this raises a question: How do foreign firms remain attractive as employers in a market that has matured past the novelty of being global?

Employer branding can’t rely on legacy prestige or foreignness anymore.

It needs to express genuine relevance to today’s workforce.
Things like clarity of purpose, empowerment, flexibility, and local authenticity consistently emerge as key factors determining younger talent’s expectations.

That means more localized decision-making, visible investment in local leadership, and a culture that feels less like a satellite of headquarters and more like an independent, confident organization in its own right.

The companies that will sustain both consumer and employee loyalty in China are those that learn to read social undercurrents.

They will adapt their narratives and structures to reflect local maturity, rather than just trying to fight against it.

In the end, the shift we’re observing may not be about brands losing consumers, but organizations losing resonance.

Whether you sell products or offer careers, the principle remains the same:

People stay where they feel seen.

And in a country as fluid and fast-evolving as China, that requires less chasing, and quite a bit more understanding.

Leave a comment

To put it brutally simple, these so-called media studios are small, nimble, seemingly autonomous groups of journalists tasked with making state propaganda more quick, palatable and, maybe most importantly, shareable.

Over the past few years, it has become quite obvious that Chinese traditional state media is simply too stiff, too slow, and sometimes disconnected with audiences.

State media are clunky, top-down systems that were built for communiqués, certainly not TikTok.

Reading about this media studios made me feel like someone at some point might have asked: What if we gave creative people smaller teams, more freedom, and a stake in their own success?

As I thought this, I found the question oddly familiar.

Haier, self-management, and the rise of the micro-enterprise

If, like me, you’ve followed Chinese management trends, you’ll see how my mind made this connection.

Haier’s RenDanHeYi model and its push for micro-enterprises is a recurring topic in business circles here in China.

The basic concept is to break a large, traditional organization into self-contained units, acting as entrepreneurial teams with broad autonomy and clear stakes in their own success.

Essentially, Haier’s model aims at reshaping giant corporate bureaucracy into an ecosystem of autonomous cells, with the stated goal of achieving enhanced agility, allowing talent to perform at their best - while maximizing responsiveness to customer’s needs and demands.

If you think about it, media studios have similar goals.
They may talk about audiences rather than customers, and be more focused on narratives instead of sales.

But I have to say, the more I think about this, the less the connection looks coincidental.

Both micro-enterprises and media studios are responses to what seems to be the same structural challenge:

How do you turn a massive, rigid organization into something that can move at today’s social media speed?

Subscribe now

Big machines, slow moves

Looking at this media studios development, it looks like the Chinese government realized what Haier’s founder confronted years back: A center that simply cannot keep up.

Chinese state media checks all the marks of traditional, slow-moving corporate hierarchical structures, with its many layers of approvals, red taping, political constraints.

The result is also very similar, with too many talented people squeezed into molds that reward loyalty, conformity and compliance over creativity.

And just like Haier, the media system seems to have realized that the solution is to reduce the weight of central planning.

You can break the machine into parts, letting them run with their own rhythm, while still tethered (more or less loosely) to the mothership.

The tether, of course, is political in the case of the media studios, while it’s economic performance in companies like Haier.

In both cases, though, this strategy is grounded on what research has showed quite consistently over the past few decades: Individuals will be more engaged and more effective when they can take ownership of their work.

But how much self-management is there, really?

Haier’s teams live or die on the market. Their ultimate goal remains the same as that of any business, regardless of how it is structured: Performance.

As I read on Chinese state media studios, on the other hand, it looks like they still live under layers of censorship and ideological training.
They may look freer than traditional state media, but the boundaries seem to remain as firm as ever.

It is Chinese state media after all.

So I wonder if we are really looking at a parallel evolution, or just a coincidence in form?

On one hand, the similarities in organizational design are striking.
On the other, the motivations and constraints couldn’t be more different.

Haier wanted to outcompete Whirlpool. The Party wants to out-narrate the BBC.

Both want to look faster and smarter.

But the scoreboard is different.

I may be seeing a connection where there’s none

Let me be clear - I’m not suggesting Beijing studied Haier case studies before green-lighting media studios.

But isn’t the convergence telling?

Large organizations (both corporations and state are large organizations after all), are facing the same structural dilemma: How do you become agile without losing control?

The answers, interestingly, tend to rhyme.

Moree and more focus on small, self-driven teams, entrepreneurial mindset as a mean to loosen up structures while tightening accountability.

There are some brilliant success stories (and some tremendous disaster ones, to be fair).

Haier is arguably a benchmark for radical organizational design, but not every firm is ready to apply the model as it is.

Will media studios work, or fail? Only time will tell.

I don’t read media studios as proof of China’s management innovations seeping into politics.

But I do see a parallel worth noticing.

When both state propagandists and global appliance makers converge on the same structural fixes, it tells us something about the pressures of operating in a hyper-speed, hyper-fragmented environment.

The problem is universal: big systems are too slow.

The solution, self-contained teams that move fast, is increasingly universal too.

The big open question is whether this model scales without collapsing under its own contradictions.

We’re living times of hyper-localization, with the China-for-China theories emerging from the ashes of the China+1 strategies, and from what my clients and contacts tell me, the biggest impact of this is in the area of decision-making.

More specifically, these senior leaders report how things have changed when it comes to decision making and the involvement of global headquarters.

In the worst stories, they tell me that decisions now simply take forever.
In the best cases, they often have to send things back up to headquarters, where the approval process gets longer and more frustrating.

What used to be a conversation with one senior decision maker has now become a multi-step negotiation with people who don’t always know the local ground.

In the words of one of these senior managers:

They keep telling us the goal is agility. Then, when we need to make fast decisions, we get stuck because someone in the US who has never even been in China, thinks we’re spending too much money on this.

You get the idea.

Subscribe now

The performance scorecard that went nowhere

Here’s a story from a manufacturing company.

The China plant leadership team wanted to update their performance management system.

They wanted to move away from rigid quarterly KPIs because they no longer reflected how work was actually done on the shop floor.

They designed and proposed a system articulated in monthly reviews, more team-based goals, and they even created some cool connections with skill development.

They thought it made much more sense locally, and they saw clear upsides with productivity.

But before implementation, it had to go back to headquarters.

After a few weeks, the first of a series of changes were requested by HQ. They wanted more benchmarks to make sure the new model would actually work.

This in turn resulted in more requests over the following week, each of them adding to the local leadership team’s frustration: More financial modeling, then a few rounds of legal reviews.

Almost three months later, by the time headquarters approved a pilot, it was simply too late.

The team had lost momentum, and it was simply too close to the end of the financial year to kick-off such an initiative.

A promising idea that had people excited for some positive and very needed change simply died under the weight of global caution.

The main problem here was that the local team didn’t frame the case in HQ’s language.

They were focusing on operational benefits, but they didn’t emphasize what HQ worries about most. Stuff like cost transparency or alignment with global standards, in this case.

What they did not have was a stronger bridge, someone able to translate both ways and influence stakeholders overseas.

This person could have kept the project alive.

Instead, a good initiative got lost in translation.

Once upon a time, there were expats

For decades, expats often played this bridging role.

Aside from the work experience, seniority, or technical expertise, expats used to play the crucial role of interpreters of context.

One of the expatriate manager’s critical functions in China was that of making this complex, vast, and oftentimes obscure land legible to headquarters.

At the same time, they would make the far-away headquarters, with their strange requests and constraints, legible to China.

The problem is that such function was useful, but it was also a crutch.

Too many companies relied on foreign managers to translate realities instead of investing in developing local leaders who could do the same with deeper legitimacy.

In recent years, expats in China have slowly disappeared from org charts - I won’t go too deep into the reasons for such a decline in numbers, but I think three years of pretty harsh anti-Covid policies and China’s resulting isolation might have played a role.

Now that many expatriates have gone, the gap is becoming (sometimes painfully) clear.

Geopolitical tensions add another layer of complexity.

Headquarters are more cautious, more risk-averse, and more demanding of “proof” before approving local decisions.

Narratives about China in many global centers are simplified, sometimes even slanted.

This makes it even harder for local voices to be trusted.

But the answer is not necessarily to go back.

I think that what international firms in China need to do, and quickly, is to build systems that develop local talent to carry that role with credibility and continuity.

Having a solid leadership bench is, after all, one of the best ways to create resilience.

But the bench is thin

This is the real weak spot.

Companies need leaders in China who can operate seamlessly between local and global.

These people understand what makes sense in Shanghai but can also explain it convincingly in Frankfurt or Chicago.

Unfortunately, that bench is not yet there. And in the few companies where it is, it is just too thin.

Chinese professionals who studied abroad help fill the gap, but this is not enough.

On the other hand, plenty of strong China-grown talent exists, but too often it has been left under-developed for global-facing roles.

Beyond the silly excuses

One argument I hear often is that Chinese managers don’t want global assignments.

Yes, some hesitate.

But the real problem, I argue, is that organizations have not invested enough in preparing them to succeed in those assignments.

After decades of leaning on expatriates, the sudden push for localization left many firms without the internal systems to properly prepare local leaders.

The gap is structural, not cultural.

Share

So what?

Building that bench as quickly as possible is the only real answer.

Among other things, it means that firms need to find that potential early, and expose their people to global dynamics.

At the same time, they need to invest - truly invest - into preparing these people to influence both local and global strategy.

It’s not a quick fix.

Building global leadership capabilities in local talent requires time, and consistent effort.

Companies that understood this five or ten years ago are in a much stronger place today.

Structures before slogans

The best organizations I know that are working on this are restructuring so that local roles more naturally connect to global processes.

A client of mine, for example, has been designing career paths that allow Chinese employees to gain international exposure early, without forcing them to choose between local and global.

They are doing this in a number of different ways, but it all started with creating cross-regional/global project teams to jointly develop selected products.

The result is a diffusion of knowledge and information across various markets - which would already be a win for many - but they’re also getting a better-connected workforce, more visibility, and transparency.

They are closing the leadership gap, and their organization is becoming a better place to work in the process.

Not bad, if you ask me.

The market stays, so must you (or not?)

Over the past few years, de-risking, decoupling, exiting the Chinese market have become a more familiar topic for discussion in more than a few boardrooms.

But for most international companies invested in China, this is unrealistic.

As another former client in the chemicals sector told me some time ago:

For almost forty years, we have invested billions into our plants here. How quickly do you think we can move them somewhere else? Decoupling is nonsense for us.

For many international firms in China the market is too important, the supply chains too integrated.

China may not be perceived as the land of opportunities it used to be when we only cared about its double-digit GDP growth and low-cost labor… but the opportunities are still here and can’t be ignored.

So the real challenge is to remain in a way that ensures decisions are localised on one hand, and globally integrated (or at least consistent) on the other.

Developing new interpreters

To sum this up, what organizations in China really need is new, influential, and credible interpreters of context.

Local teams know what makes sense, yet their message stalls on its way to headquarters.

Expatriates used to fill that gap, but for many reasons they may no longer be the solution.

The real fix lies in building a deeper leadership bench, one that can bridge local and global with credibility and continuity.

Until companies invest in those systems of knowledge diffusion and talent development, they will keep losing ideas, momentum, and trust in translation.

One thing I often ask to leaders faced with this issue is: If your local team had a great idea today, would it make it through the round trip? Would it survive the journey to, and back from, headquarters?