The meeting your dashboard cannot see: the place where staying or leaving is quietly decided.
The exit interview takes twenty-two minutes. It is the fourteenth this quarter, and the HR business partner conducting it could run it in her sleep. Across the table sits a project manager: eight years with the firm, three towers' worth of delivered infrastructure, two promotions, genuinely liked. Between them, on a screen, is a form with a dropdown menu. Reason for leaving. He studies the options the way you study a menu in a restaurant you didn't choose. Compensation. Career progression. Relocation. Personal reasons.
He selects career progression. She thanks him, warmly and sincerely. Both of them are relieved. The form is filed, the dashboard refreshes, and somewhere upstairs a quarterly people report acquires one more data point in a bar chart labelled with the most soothing phrase in human resources: market forces.
Forty minutes later, in the car park, he tells a colleague the truth. It comes out in fragments, the way true things do. Feedback from his manager that landed, for three years, like sandpaper: delivered briskly, publicly, in a style his manager sincerely believed was "just being direct." Meetings paced so that the beat of silence he needed before disagreeing was read, every time, as having nothing to say. The slow discovery that being excellent and being heard were, in this company, two separate careers, and only one of them was his.
None of that is on the form. There is no dropdown option for 'My manager never learned how I disagree'.
I want to persuade you of something about that twenty-two-minute meeting: it was the wrong meeting, held with the right man, at the wrong time. And your organisation is almost certainly holding the same meeting, this quarter, at scale. To see why, you need to know when he actually resigned.
"There is no dropdown option for 'My manager never learned how I disagree'."
Every resignation happens twice
Here is the observation at the centre of this piece, and once you see it you will not stop seeing it: every resignation happens twice. There is the resignation your systems record: a letter, a notice period, a dropdown, a leaving card. And there is the first resignation: the private one, months earlier, unwitnessed and unminuted, when a person quietly concludes that the future they wanted does not live at your company. Everything after that moment (the updated CV, the coffee with the recruiter, the sudden flexibility about "exploring options") is not a decision. It is logistics.
Our project manager's first resignation has a date. It was a Tuesday in March, fourteen months before the exit interview. A planning meeting, mid-morning: he had spent the weekend building a phasing proposal that would have saved the client six weeks. He began to present it. He was talked over, not maliciously, just fluently, and when the room paused and he took the breath his communication culture takes before contradicting a senior colleague, the chair read the pause as agreement and moved on. Twenty minutes later a colleague restated half of his idea at twice the volume and collected the credit. Nobody in that room behaved badly by their own lights. And a man who had given the firm eight years walked out of it having signed nothing, said nothing, and decided everything.
No system on earth recorded that Tuesday. The dashboard met his resignation fourteen months later and called it career progression, which, by then, was even true. That is the maddening elegance of the thing: by the time the second resignation arrives, the first has dressed itself in a reason nobody can argue with.
Between the two resignations lies what I've come to call the ghost quarter, often two of them, sometimes more: the interval in which a person has already left and is simply still attending. They are polite in the ghost quarter. They deliver. They also stop volunteering, stop challenging, stop making the small bids for connection that colleagues make when they believe in a shared future. Their disengagement is quietly pricing itself into your projects long before it prices itself into your attrition rate. Gallup's exit research contains a devastating detail here: roughly half of departing employees say that in their final three months, no manager or leader spoke with them about their satisfaction or their future at all. Read that alongside the two-resignation structure and it stops being a statistic about negligent managers. It is a statistic about timing. The conversations are happening, but around the second resignation, when there is nothing left to discuss. The first resignation, the only one that was ever negotiable, passes in silence. Which is why the same research finds that around half of all voluntary leavers (three-quarters, in the Work Institute's exit studies) say their departure could have been prevented. Prevented, that is, by someone who knew which resignation to attend.
People rarely resign from the company. They resign from a hundred small Tuesdays.
"People rarely resign from the company. They resign from a hundred small Tuesdays."
The four costumes
If the first resignation is invisible, what exactly is your attrition data looking at? It is looking at costumes. The dropdown on the exit form offers four of them, and cultural friction, which has no field of its own, leaves the building wearing whichever one fits best. Consider this a field guide.
Compensation. The unarguable alibi. Salary is a wonderful reason to give: it is objective, it flatters the employer (we loved it here; they simply paid more), and it demands that nobody change anything. It is also, frequently, true, in the trivial sense that the person did, in fact, accept a better-paid job. But notice what the costume conceals. An employee who feels read, backed and heard entertains the recruiter's call politely and stays; an employee in his ghost quarter takes the call in the stairwell with his heart rate up. The offer didn't cause the departure. The first resignation caused the offer to be listened to. This is why counter-offers fail at the rates they famously do: you cannot counter-offer someone out of being unseen. Money answers the question the form asked. It has no purchase on the question the form didn't.
Career progression. Regular readers of these Field Notes have met this costume before, in the Emiratisation files: "career development" is what it says on the form; there was no path here for someone like me is what it means. It generalises far beyond Emirati talent. In a workplace running on one culture's rules for how ambition must be performed (self-promotion at appraisal, elbow-first competition for the stretch project, fluency in the meeting's native cadence), everyone who signals differently watches a promotion system that calls itself meritocratic reward one very culturally specific style of merit-display. They draw the rational conclusion: the ceiling here is real, and it is made of other people's norms. Then they leave "for progression," and the word does the laundering.
Relocation. The rescue that was requested. The spouse's opportunity abroad, the ageing parent, the schooling decision: all real, all human, and all, curiously often, acted upon in the ghost quarter. Families weigh relocations continuously; what tips the scales is rarely the pull of elsewhere and usually the absence of a countervailing pull from here. A person anchored by belonging finds reasons the move can wait. A person who resigned privately in March finds the spouse's offer arriving in June "impossible to turn down." The taxi to the airport was booked, in truth, months before the flight.
Personal reasons. The costume of last resort, where the truth goes to be respected and ignored. It is the category chosen by the people for whom the real answer is most cultural and least sayable: the manager whose "banter" had a target, the team whose socialising ran through venues she couldn't join, the daily arithmetic of being the only person in the room paying what I have elsewhere called the fluency tax. "Personal reasons" is the form's way of agreeing not to ask. Everyone honours the treaty. The file closes.
Four costumes, one wardrobe. And note what they share: each is true enough to survive an exit interview, and each is downstream of a first resignation the form was never designed to detect. This is what I mean when I say that attrition data arrives on the executive floor pre-laundered. Cultural friction leaves your building every single week, dressed, respectably, as compensation.
The arithmetic
Now let us price the laundering, because your board will want numbers, and the numbers are genuinely violent. Gallup puts the cost of voluntary turnover to US businesses at around a trillion dollars a year, and the cost of replacing one employee at between half and twice their annual salary: nearer 40 per cent for frontline roles, around 80 per cent for technical professionals, and fully 200 per cent for leaders and managers, once you count recruitment, onboarding, the productivity trough, and the slow evaporation of everything they knew that was never written down.
Then add the Gulf's own multipliers, because ours is not a generic labour market. Here, the employer often carries the visa, the flights, sometimes the housing, an expatriate premium on every hire. Here, roughly nine in ten private-sector employees came from somewhere else, which means essentially all of your attrition is cross-cultural attrition, whether your dashboard has a column for that or not. And here, above all, is a relationship-first market: trust is built person to person, over years, in majlis and lunches and a hundred small reliabilities. It belongs to people, not to institutions. When a client-facing leader resigns, a measurable share of your client's trust gets into the same taxi.
Run those figures across a two-hundred-person firm losing a sixth of its people a year, and the question changes shape. It stops being can we afford to address cultural friction? It becomes: how long can we afford to keep paying for it without once writing its name down?
Autopsy and epidemiology
Why does an intelligent organisation, staring at numbers that size, keep misdiagnosing them? Because of when it looks. Your attrition dashboard is a coroner: rigorous, precise, and only ever introduced to the departure after the fact. The exit interview is worse: an autopsy conducted with the chief witness already halfway out of the door and every incentive to be kind. The departing employee needs a reference, sometimes a visa transfer, always an unburned bridge in a market where everyone knows everyone. The interviewer needs the meeting to end without incident. Harvard's major study of exit-interview programmes found what these incentives predict: nearly everyone collects the data; almost nobody can point to a decision it changed.
What autopsies are to medicine, exit data is to retention: necessary, and structurally incapable of saving the patient on the table. The discipline you actually need is epidemiology: the study of the living. And here is the good news buried in everything above: the evidence is already in your possession. Two investigations, neither requiring a dirham of new software.
First, run the one segmentation almost nobody runs: regretted attrition, broken down by cultural group, set against the cultural composition of the leadership each group reports into. Where the same managers show meaningfully different retention across cultural groups (and in my experience of Gulf organisations, they almost always do), you are not looking at a pay problem or a market problem. You are looking at a reading problem: leaders fluent with people like themselves and illegible to everyone else. That pattern has been sitting inside your HR system for years, waiting for someone to ask. Segment the data honestly and it will confess.
Second, measure the gap between confidence and experience. Ask your leadership team to rate its own cultural intelligence; then ask it to estimate how colleagues from cultural minorities would rate it. Decades of research on leadership self-assessment predicts the result: the first number reliably exceeds the second. And notice that both numbers are still the leadership team's own. Almost no organisation has ever measured the second number directly, by asking the colleagues themselves. In the space between the estimate and the reality, next year's first resignations are being decided right now.
"Segment the data honestly and it will confess."
The organisations that see it
The firms that get ahead of this (and having watched a few closely, they are rarely the ones with the largest inclusion budgets) make three unglamorous moves.
Structured, genuinely safe listening across cultural groups, conducted while people still have a future to discuss, with a visible pre-commitment to act on what comes back. One well-run listening exercise, honoured, is worth fifty immaculate autopsies, because it meets people on the near side of the first resignation, while the decision is still a conversation and not yet a costume.
Not another heritage day: the specific, learnable leadership craft of surfacing friction early and resolving it in ways that preserve dignity on every side, because friction that goes quiet does not dissipate. It accrues interest, in the currency of Tuesdays, and it compounds towards a March meeting nobody will remember and everybody will pay for.
Written into what it takes to lead here, assessed like commercial acumen: can this manager read communication styles other than his own, hear the pause before the disagreement, run a meeting in which more than one dialect can score points? Whatever appears in promotion criteria is the only values statement employees fully believe. Everything else is a poster. (You may sense a consultant's interest declared in all three of these; I declare it. The arithmetic stands anyway.)
The line item that does not exist
One floor up from that exit interview, some weeks later, the quarterly people report is on the screen. Attrition: discussed. Pipeline: discussed. Compensation benchmarking: commissioned, again. It is the standard remedy for a laundered diagnosis, and about as effective as re-pricing the menu in a restaurant people are leaving because of the conversation at the table. At no point does anyone ask the only question this article has been circling, because no template on the screen forces it: what share of our attrition is cultural, and what is it costing us?
It is an answerable question. Given the sums involved, it may be the highest-value answerable question in your people data. But no dashboard displays it, no dropdown collects it, and so it goes unasked, quarter after quarter, taxi after taxi.
As for our project manager: his first resignation had a date, a room, and a sentence that was never finished. Fourteen months of notice, in effect: the longest notice period in the company's history, served in plain sight, attended by no one. The exit interview took twenty-two minutes and recorded that he left for his career.
He didn't. He left on a Tuesday in March, from a meeting that is still, somewhere in your building, being held.
"Fourteen months of notice, served in plain sight, attended by no one."