One in five.
That is the global engagement number in Gallup’s 2026 survey.
One hundred and fifty thousand people. One hundred and sixty countries. Probably the most disciplined long-term measurement of workforce attention, energy, and withdrawal we have.
Twenty percent engaged.
The rest are either dragging themselves through the system or have already left emotionally. They still have a badge. They no longer have a stake.
The lowest reading in years.
At the same time, one of the largest coordinated capital reallocations in corporate history is moving toward AI. BCG and the World Federation of People Management Associations spoke with more than 7,000 leaders across 115 markets this March. Strategic workforce planning moved to number two. Digital HR automation jumped 13 places in three years, the largest move in the history of the survey. Every AI-adjacent topic moved upward.
Two readings of the same institution.
One says the human layer is detaching.
The other says the machine layer is about to be loaded with more ambition, more capital, and more board attention.
The AI program sits on top of engagement.
Right now, that floor is cracking.
/////
The BCG report does not say engagement no longer matters. That would be an unfair charge. It says engagement and employee experience have become “foundational rather than differentiating.”
Reasonable on the surface.
Also the perfect description of what gets eaten first when capital is reallocated.
Foundational things don’t win quarterly comparisons. Differentiating things do. If you say “both matter” without deciding which one carries the other, the system will decide for you.
It always does.
Who benefits from the current arrangement?
Executives running AI programs have a clean timetable. By the next board meeting, they need to show ROI. Adoption rates. Automation savings. Integration milestones. They can produce a number this quarter.
Engagement does not behave like that.
Nobody loses his job because engagement drops three points in Q2. Plenty of people get shouted at because an AI milestone is missed in Q3.
In BCG’s ranking, engagement and wellbeing fell seven places between 2023 and 2026. Rewards and recognition fell six. Once a new board-legible category enters the room, the machinery knows where to look.
This time the category is called AI.
/////
For many years I thought transformation programs failed for the standard consultant reasons.
Wrong tool. Wrong sequencing. Insufficient sponsorship.
Then three years would pass, and I would find myself in another room, with the same logos, listening to the third version of the same transformation. The post-mortem always blamed the tools.
The tools were usually not the problem.
The missing piece was simpler and more brutal. The people responsible for execution had mentally resigned before the change arrived.
Gallup’s data makes the pattern visible. Manager engagement fell from 31 percent to 22 percent in three years. Gallup attributes most of the variance in team engagement to the manager. The broader literature is more cautious, usually somewhere in the 30 to 50 percent range.
Choose your number. The message does not change.
The line manager is the closest lever the organization has to the actual work. That lever has been quietly bending. The standard institutional response has been to put more weight on it.
This collapse has at least three readings.
The first is the one I act on. The manager role has been overloaded by decades of administrative sediment, and AI programs have landed on top of an already saturated function. Clean the role, and the rest has a chance.
The second is harder for the manager class to hear. Some managerial work really is substitutable by AI. Capital allocators may be sensing this before the substitution is complete. Restoring capacity does not remove the underlying signal.
The third is macro. AI-related job anxiety, post-pandemic stress, political instability across continents, climate anxiety sitting underneath the whole thing. If this is the dominant force, a role audit may move the line a little, but it will not reverse the trend.
I act on the first reading because there is a lever there.
Not because I am sure it is the whole truth.
The honest position is less comfortable: all three may be partly true at the same time.
/////
The manager role has accumulated sediment.
Compliance files. Escalation chains. Performance documentation nobody reads. Meetings nobody has the authority to cancel. Dashboards created because one senior person once asked one nervous question in one bad quarter.
The late Ottoman bureaucracy worked with a similar logic. Every reform added a layer. Almost no reform removed one. By the late nineteenth century, the Tanzimat reformers had to use the very institution that made reform difficult.
We built measurement architecture that made engagement invisible at board level, and then we called this prioritization.
Board packs rarely show three-year manager engagement decline with a named owner attached. Executive compensation rarely punishes consecutive engagement deterioration. The system was not designed as negligence. That would be too flattering. Negligence at least implies somebody noticed.
We inherited a governance structure built before engagement became a serious category, then never asked whether we would build the same structure today.
We are not outside this.
We built it.
We maintain it.
Most of us benefit from it, most of the time.
/////
The loop is clean.
AI programs land on managers who are already overloaded. Execution quality drops. Senior leadership tightens deadlines. The manager’s cognitive margin shrinks. Team engagement falls. Pressure increases.
Then the same system asks the same manager to carry more transformation.
Cheap cheese sits in an expensive mousetrap. In this case the cheese is called productivity.
MIT’s Project NANDA reported that 95 percent of organizations deploying generative AI saw no measurable return. The thresholds and time windows deserve scrutiny. Some criticism is fair. But no serious adjacent reading says, “Actually, everything is going beautifully.”
The direction is the direction.
If manager capacity is collapsing this badly, why don’t strong managers simply leave for organizations that have designed the role better?
Some do.
Most don’t, because the same overload has spread almost everywhere at once.
This is mainly a large-company disease. In BCG’s data, smaller firms still place culture, rewards, and skills closer to the center. The argument here applies most directly to multinationals, listed companies, and private equity portfolio businesses.
Partner-led firms, founder-led companies, and family businesses suffer from the same world through different pathologies.
Their ghosts have different surnames.
/////
The prescription is not pleasant.
The manager role needs a zero-based audit.
Not a wellness program.
Not another leadership module.
An audit.
What is inside the role today? If we designed the role from scratch, what would we never put there? Which reports exist only because nobody had the courage to kill them? Which approvals are fake control? Which meetings are theatre? Which staff functions have outsourced their friction to the manager and called it process discipline?
Some of this work has been automatable for years.
Some of it exists because a function needs to justify headcount.
Some of it is compliance theatre for a regulator who may not read what we produce.
This is not a new idea. It exists in the consultant playbook. BCG’s “Four Power Moves” points in broadly the same direction. My disagreement is not with the components. It is with the order.
Restore capacity first.
Then add.
Most large organizations are doing the opposite. This is why every round produces the same exhaustion with a better slide deck.
Who should run the audit?
Not a committee.
Committees don’t eliminate their own inputs.
It has to be led by an executive whose performance review does not depend on preserving the current architecture. On paper, the obvious candidate is the CHRO. In practice, that is exactly why it usually fails. I have watched CHROs enter boardrooms with the right diagnosis and leave outvoted by colleagues whose KPIs are quarterly and whose patience is shorter than the problem.
/////
Late-cycle institutions rarely redesign themselves under strategy.
They redesign under crisis.
I am working on this more deeply in my forthcoming book.
Ibn Khaldun, writing in the fourteenth century about the life cycle of civilizations, would have recognized the pattern. He called the binding force asabiyyah. Institutions start with shared purpose and end with process compliance. The first generation has a mission. The fourth generation has a template.
We are not far from that.
Twenty percent engagement is not necessarily a warning before the threshold. For many organizations it may already be the steady state. The remaining eighty percent have made a calculation. Discretionary effort costs more than the system returns.
This is not laziness.
It is arithmetic.
That calculation changes only when something forces it to change.
A talent shock. A competitor that solves the constraint and takes three points of market share. A board that misses AI ROI for two consecutive quarters and finally asks why the same technology works elsewhere but not here.
Organizations treating manager capacity as a background variable while loading it with AI requirements are not executing a strategy.
They are placing a bet.
The bet is simple: the floor will hold long enough for the AI investment to pay back before the cracks reach the beams.
Some floors hold longer than you expect.
Some collapse without giving a speech.
I wrote this article in Turkish too. They share a skeleton. They don’t share skin, muscle, or voice.
You can read it here.
https://hguvenal.substack.com/publish/post/196406388?r=nslov&utm_campaign=post&utm_medium=web&showWelcomeOnShare=true