The retention problem is a selection problem you didn’t notice

When a senior leader resigns, the response is almost always present-tense: what changed recently, what can we counter with, how do we make them stay. But a senior departure is rarely a recent event. The seeds were usually planted at selection: a mismatch between the leader and the actual mandate, a promise the role couldn’t keep, a misread of what would keep this specific person engaged. By the time they’re leaving, that mistake has been quietly compounding for a year and a half. Senior churn isn’t rare, ICRA puts top-tier IT attrition near 13% and Aon overall India attrition at 16.2%, so the cost of a selection error, multiplied across a bench, adds up fast.

This reframes the whole problem. Retention is largely determined by whether you put the right person in the right seat for the right reasons in the first place. The exit interview is the end of a retention conversation you didn’t know you were having.

Most of what centres label retention is triage: the counter-offer, the retention bonus, the hurried conversation about growth once someone’s already halfway out. Sometimes it works for a quarter. It rarely fixes the underlying mismatch, because the mismatch was baked in at selection and money doesn’t un-bake it. As Harvard Business Review argues, a counter-offer fixes the compensation and leaves the reason the person was leaving untouched. You’ve paid to delay a departure whose real cause you never addressed, and often just moved the exit a few months down the road.

The leaders who stay in pharma GCCs, where the scarce specialist bench is hard to rebuild, tend to be the ones who were correctly selected: whose mandate matched what they actually wanted to do, whose sponsor was real, whose role was described honestly. India’s life-sciences GCCs have deepened into roughly 45% of drug-discovery and 60% of regulatory work, on EY’s read, so the specialists worth keeping are the hardest to replace. Retention was a property of a good decision made before they signed.

Look closely at a selection error and it’s rarely incompetence. The person is good. They were good at the job you interviewed them for, which turned out to be a different job from the one that existed. A data-engineering lead hired to build gets handed a mandate to maintain. A functional head sold on autonomy reports into a parent-company matrix that decides everything for them. Their skills matched the spec. The situation didn’t match the story. And situation is the part an interview is worst at surfacing, because everyone in the room wants the meeting to go well.

You can test fit-to-mandate if you decide to. It means writing the mandate down honestly, including the 18 months of unglamorous work nobody puts in a job spec, and then checking whether this person actually wants that work in its real form. It means asking what would make them leave early, while they can still answer without a resignation letter in their hand. Selection done this way treats the offer as a hypothesis about a 2-year fit, and gathers the evidence to test it before the offer goes out. That’s the difference between a panel that liked someone and a panel that understood them.

I come at this like a systems problem. When something fails 18 months downstream, you don’t only ask what broke last week. You trace it back to where the tolerance was set. A senior resignation is feedback, and the honest version of that feedback loop runs all the way back to the selection panel: what did we believe about this person, and where were we wrong? Most centres close the loop at the exit interview, which sits far too late in the pipe to change the thing that actually broke.

The uncomfortable part is that the panel is usually well-meaning and sure of itself. That certainty is the problem. A candidate who interviews beautifully produces a feeling of confidence that has very little to do with how they’ll hold up in the real mandate, and the gap between the two is where selection error lives. You can shrink it. Score people against the mandate you wrote down, put the person who owns the outcome in the room, weight evidence of judgment above the warmth of a good conversation. None of that is exotic. It’s just rarely treated as the retention work it plainly is.

So the highest-return retention work is unglamorous and early: a sharper selection decision. Be honest about the mandate, including the parts that don’t sell. Test genuine fit-to-mandate, and see past the comfortable fit-to-vibe that flatters the panel. Understand what specifically will keep this individual engaged two years out, and select for that ahead of who interviews best. Every bit of rigour you add here is retention you never have to buy later.

None of this shows up in a retention dashboard, which is exactly why it’s under-invested. The dashboard measures departures after they happen. The work that prevents them happened before there was anything to measure, and only about a quarter of talent leaders feel confident they can even measure quality of hire, per LinkedIn, which is why the selection decision stays under-examined. If your centre keeps losing leaders around the 18-month exit cliff, go look at the selection decision. That’s where the problem was always hiding.