Retention is a sourcing problem

When a strong hire leaves at 18 months, the retention post-mortem usually starts too late. It looks at the manager, the comp review that came in flat, the reorg, the project that got shelved. Those are real, and sometimes they're the whole story. More often they're the surface. Underneath sits a gap that opened on day one: the job the person accepted and the job they walked into never fully matched.

The mismatch is rarely a lie. It's the accumulated optimism of a search. The scope was pitched at its widest, the mandate implied rather than fixed, the growth path described as if it were already funded. None of it was false when it was said. It just wasn't load-bearing, and the candidate built their decision on it anyway.

The cost of getting this wrong isn't small. Replacing a mid-to-senior professional runs to a meaningful share of annual salary once you count the search, the ramp, and the work that stalls while the seat is empty. And the exit clusters where it hurts: SHRM and wider attrition research consistently show the sharpest voluntary departures land in the first 2 years, before the hire has returned the investment made to bring them in. A retention programme aimed at year three is guarding the wrong door.

Nobody sets out to mis-sell a role. The incentive does it for them. A recruiter working a hard requisition is measured on the close, and the fastest path to a close is to describe the job at its most attractive. The autonomy that exists on a good week becomes the autonomy of the role. The stretch project that might land becomes the mandate. The promotion that leadership gestured at becomes a timeline in the candidate's head.

The candidate does the rest of the damage themselves. Under offer pressure, people hear the ceiling as the floor. “You'd likely own the platform strategy” is stored as “I will own the platform strategy.” “There's room to grow into a lead role” is stored as a promotion date. Nobody wrote anything false. Two people just left the conversation holding different versions of the same job.

Then the reference-checking of expectations never happens. We verify a candidate's track record with forensic care and verify their understanding of the role with none. The offer goes out, gets signed, and the gap between the pitched job and the real job goes into the background, where it compounds quietly until the person is close enough to the work to see it.

A calibrated pitch is a specific one. The difference is that every claim in it is something the hiring manager will still stand behind in the person's second quarter, when the honeymoon lighting is gone and the work is just the work.

That means separating what's committed from what's possible, out loud. The committed part is the scope in the first 6 months, the reporting line, the budget that exists today, the decisions this role actually gets to make. The possible part, the expansion of scope, the team that might get built, the promotion that depends on a headcount plan not yet signed, gets named as possible, with the condition attached. “There is a path to leading a team; it depends on the function hitting its number and a headcount approval we don't have yet” is a stronger sentence than “you'll build a team,” because it survives contact with year two.

It also means being straight about the parts that are hard. The stakeholder who's difficult. The system that's held together with duct tape. The 2 quarters of unglamorous cleanup before the interesting mandate begins. Candidates worth hiring accept an honest hard problem; they flee a hard problem they were told was easy. The calibrated pitch costs a few closes with people who were only ever buying the fantasy. Those were the year-two exits you were going to eat anyway.

The other half of the fix is on the assessment side. Most interview loops are built to find out whether the candidate can do the impressive version of the job. Fewer are built to find out whether they'll still want it when it's the ordinary version. Those are different questions, and the second one is the retention question.

It's answerable in the room. Walk the candidate through a realistic first quarter, the actual first project, the actual constraints, the actual state of the systems, and watch which way they lean. Some light up at the cleanup because they've done it before and know it's where the real gains are. Some visibly deflate, because the job they were interviewing for was the keynote version. The deflation isn't a reason to reject; it's a reason to keep talking until the real job and their real appetite are the same job or clearly aren't.

Ask what would make them leave. Ask what the last role promised that it didn't deliver, and listen for whether the disappointment was about the work or about a story they were told. A candidate who can name the gap in their last job is a candidate who'll notice you closing it in this one. That's the fit-to-reality screen: whether the bar they clear will still hold their attention 18 months from now.

Reframing retention as a front-end problem moves the budget. The engagement survey, the stay interview, the manager-training module all have their place, but they're downstream of a decision that was already made in the offer conversation. Money spent tightening the pitch and the fit-screen buys expectations that hold; money spent on retention programmes buys the repair of expectations that didn't.

For a GCC or an enterprise TA function, the practical shift is to treat the offer stage as a retention control point and instrument it accordingly. Hold the recruiter and the hiring manager to the same standard of accuracy on the role that you hold the candidate to on their history. Write the committed scope down. Name the conditions on the growth path in the offer conversation, not after. It's cheaper to set an expectation correctly once than to manage a disappointed high performer through the exit they'd half-decided on by month 12.