The best people leave first. Here’s why that’s predictable.

When a role disappoints, everyone in it feels the same friction. Not everyone can act on it. The strongest people in a function have the most doors open to them and the shortest fuse for a seat that isn't what it claimed to be. They simply have somewhere to go and no reason to wait for the situation to improve on its own.

The median performer sits in a quieter market. The options are fewer, the recruiter calls are softer, so they stay through the exact same friction that moved the best people out. Read one resignation at a time, every exit looks like a personal decision with its own story. Read the pattern across a team over a year, and it's a sorting mechanism. It sorts from the top, because the top is who the market is bidding for.

This is why the outcome is predictable even when the timing isn't. You may not know which of your strong people will move this quarter. You can be fairly sure that if the seat under-delivers, the ones who move will be drawn from the group you least wanted to lose. Attrition isn't neutral about who it takes.

A centre can post a comfortable attrition figure and still be hollowing out, because the number counts heads and not weight. Losing the handful of people who anchored the hardest work does more damage than a percentage point implies, and the damage is invisible until a critical deliverable stalls and someone asks who used to carry it. By then the answer has already left.

So the aggregate reassures you at exactly the wrong moment. A team can shed its ceiling and keep its floor while the dashboard stays green, because the dashboard was never built to weigh the people it counts. The question that matters is whether the ones who left were the ones holding the function up.

The defence is making sure the seat your strongest people occupy actually pays out on its promise, so that having options doesn't automatically mean using them. Retain the ceiling on purpose, because the market will always try to take it first, and the reason it usually succeeds has less to do with pay than most exit interviews admit.

Comp is the socially safe thing to cite when you leave. It doesn't burn a bridge, it doesn't name a person, and it's hard to argue with. Nobody interrogates someone who says they got a better offer. So it becomes the default line in the exit interview, and organisations, hearing it repeated across enough departures, conclude they have a pay problem and open the compensation review.

Often the real trigger is the manager, and pay is just the reason that's easy to say. What wore them down was a weekly one-to-one that kept getting cancelled, a scope that kept getting clipped, a leader who couldn't or wouldn't unblock them. The offer letter from elsewhere was only the mechanism; the reason had been building for months in the seat they were sitting in.

The tell is in the clustering. When exits pool under one or two managers while the rest of the function holds steady, no comp band explains it, because the pay structure is the same across the team and the retention isn't. That's a management signal dressed as a market signal. Diagnose it as pay and every lever you pull misses: you correct comp for a team that wasn't underpaid, the departures continue, and you've spent the budget and kept the cause. Diagnose it right and the fix is uncomfortable but cheap. It's a conversation with a manager.

When we look at where senior departures cluster, they don't land in the first year as often as people assume. The first year is busy. The hire is learning the organisation, building relationships, establishing standing, and there's enough novelty and enough goodwill to carry a role even when its scope turns out thinner than advertised. The hire gives it the benefit of the doubt, tells themselves the mandate will grow, and gets on with proving they belong.

The second year is where the story turns. By then they know the place, they've done what the role asked, and they look up expecting the mandate to have expanded to meet them. Often it hasn't. The title is the same, the authority is the same, and the ceiling that was invisible in year one is now plainly overhead. That's when the recruiter's call gets a different answer. The person outgrew a mandate that was never designed to grow. It's the same scope problem from the exit interview, only now you can see the clock it runs on.

The centres that hold their senior people past the 2-year mark treat the mandate as something that has to expand deliberately. They plan for the leader to own more by year two than they did at the start, a bigger problem, a wider remit, a decision that was previously above them, and they build it into how the role is set up. The signal that fails is the opposite: a senior person who has to ask for more scope and gets a maybe. By then the calculation has already shifted, and the recruiter who called last month is looking better than they did.

The stay interview is a genuinely good idea. Ask a valued senior why they'd stay while they're still staying, and you get the signal before it hardens into a decision. The problem is how it's usually operationalised: handed to HR, scripted, logged, and disconnected from anyone with the authority to change the thing the person names. Asked by someone who can't act, the honest answer never comes, because why spend it? So the conversation collects safe, generic replies, gets rolled into an engagement number, and confirms that everything is broadly fine right up until the resignation says otherwise. The instrument works. The circuitry around it doesn't.

When it's run instead by a manager who can actually move something, and who has earned enough trust to get a straight answer, the reason that surfaces is scope. The role has narrowed, the ownership has drifted, the interesting work went somewhere else. Which is exactly the year-two break, caught a few months earlier while it's still a problem you can solve. No survey instrument catches this, because no survey asks the question that would, and no engagement score is built to register a ceiling.

So the fix is a plumbing fix. Put the conversation with the person who owns the seat, hold it early, and make sure they can act on what they hear. Done that way, the stay interview becomes an early read on the one thing that actually moves your strongest people. Done the other way, it's a survey that reassures you until the day it can't.

The through-line across all of it is that senior attrition is legible once you read the position each strong person sits in. The strongest people leave first because they can. They leave the seat and the manager, then name a number because a number is safe. The break lands in year two when a static mandate finally reads as a ceiling. And the one conversation that could surface it early only works when the person asking has the authority to fix what they hear.

The fix is to weigh exits instead of counting them, to look at which managers the departures cluster under, to grow the mandate on purpose before a senior hire has to ask, and to move the stay interview from a form to a conversation the manager owns. If your strongest people are the ones the market takes first, the useful response is to make sure their seat pays out on its promise before someone else offers them a better one.