The Offer Fall-Off Risk Read for Late-Stage Candidates

You have a candidate at offer stage and one question that decides your next move: how likely are they to renege or take a counteroffer? This guide is for in-house recruiters, sourcers, talent leaders, and founders doing their own hiring. It gives you a repeatable way to grade the specific candidate in front of you as low, medium, or high fall-off risk across named dimensions, then translate that grade into a securing action and a backup decision. It is a scoring framework, not a list of keep-warm tips.

Why fall-off deserves a score, not a checklist

Fall-off is common enough that guessing is expensive. A Robert Half survey found 28% of candidates accepted a job offer and then backed out, and a Gartner study of nearly 3,500 respondents is cited two ways: 35% reneged on accepted offers, and, in another framing, 50% accepted an offer, backed out, and started elsewhere. Interns are trending the same direction: NACE members reported renege rates rising from four to five percent to over eight percent.

The problem with most advice is that it treats reneging as one undifferentiated risk you fix with better communication. But the candidate who has three inbound offers and cited pay as their reason for leaving is nothing like the candidate who took a pay cut to join a mission they have followed for years. A single dial ("keep them warm") applied to both wastes effort on the safe one and under-secures the exposed one. A score lets you spend your securing energy where it changes the outcome.

The other reason to score: fall-off is not free. The Work Institute puts the cost of a voluntary exit at roughly 33% of base salary, and a renege lands you back at square one with a role you thought was closed. A five-minute risk read that catches even one exposed offer pays for itself many times over.

The dimensions that predict fall-off

Score six dimensions. Two are biographical (what the profile says), one is behavioural (what the candidate does), and three are situational (the market and economics around the offer). Weight the behavioural and situational above the biographical, because that is where the validated predictors live.

A comprehensive meta-analysis by Rubenstein and colleagues identified 17 validated predictors of voluntary turnover, and the strongest ones - intent to quit, job satisfaction, commitment, and job search behaviour - are behavioural, not static. Only two of the 17 typically live in core HR systems at all. That is the single most important fact in this guide: what a candidate does tells you more than what their resume shows.

Biographical signals: use as context, never verdict

Prior tenure is the classic trap. Research from the Cornell Center for Advanced Human Resource Studies found previous tenure predicts future tenure with the same employer at a correlation of only 0.22. That explains almost none of the variance. A staff engineer who held three roles in five years is not automatically a flight risk, and rejecting or over-securing them on that basis is a false positive you pay for.

Behavioural signals: the loudest alarm you have

After saying yes, 47% of candidates report they would remain open to a late arrival offer. So a candidate who tells you "I'm still weighing another process" or who has a fresh public "open to work" footprint is signalling louder than any CV gap. Documented reneging reasons back this up: 44% left for a better offer, 27% for an acceptable counteroffer, and 19% because they heard bad things about the hiring company.

Skill scarcity is a competing-offer multiplier you can measure

The scarcer the profile, the more inbound and counteroffers it attracts, which raises fall-off risk directly. You can quantify this rather than guess at it. In Refolk's index of professional profiles, the segment counts make the point concrete.

A candidate who sits in that 0.17% Rust slice will field more inbound and be harder for their current employer to replace, which makes a retention counteroffer more likely. Scarcity belongs in the risk model as a direct upward adjustment, not as a footnote on the profile.

To read scarcity before you make the offer, you need to see how thin the candidate's slice actually is against the broader pool. Refolk turns that into a plain-English search rather than a manual count.

Geography is a scarcity cousin. Counteroffers cluster in hot metros: the Financial Times reported 40% of employers made a counteroffer to retain a leaver, but in London that figure was 58%. The same candidate is materially harder to secure in a hot market, independent of anything on their profile, so location earns a place in the model too.

Counteroffer economics: model the counter before you call it secured

Assume any strong performer may be counteroffered, because roughly 50% of candidates who resign receive one. The question is not whether the counter comes but whether it works, and that turns almost entirely on the candidate's real motive for leaving.

CEB data shows 50% of employees who accept a counteroffer leave within 12 months. That is the durable risk: the counter papers over a pay gap, but if pay was the only motive, the same dissatisfaction resurfaces. Your job at offer stage is to find out whether pay is the primary driver.

If the candidate cited pay as the primary reason for leaving, rate counteroffer vulnerability high and check whether your offer clears the likely counter with margin. If they cited scope, manager, mission, or growth, the counter has less to grab onto, and vulnerability drops even for a well-paid candidate.

The counter almost always comes for a strong performer; whether it lands depends on whether pay was the only reason they were leaving.

Turn six dimensions into one score, with the driver named

Collapse the dimensions into a single low, medium, or high grade, and name the one dimension that drives it. The point of naming the driver is that your securing action attaches to it: a high score driven by counteroffer economics needs a different fix than one driven by an active job search.

kind: matrix
title: Fall-off risk from behaviour and situation
caption: Behavioural intent on one axis, situational pull on the other, decides how hard to secure.
x: Off-market, not shopping :: Actively shopping
y: Employer unlikely to counter :: Employer will fight to keep
quadrant: Low risk: standard monthly keep-warm :: Medium risk: hiring-manager call, faster paperwork
quadrant: Medium risk: close the economic gap early :: High risk: accelerate start, close gap, keep backup warm

The scoring rule that matters most: a single high behavioural signal outweighs a clean tenure record. If a candidate says they are still open, or their responses slow and cool, that pushes the composite to at least medium regardless of how stable their history looks. Score conservatively when behaviour and biography disagree.

Benchmark your read against segment acceptance rates so you know when a candidate is in a naturally leakier band. Niche and software roles already sit lower than standardized ones.

A software engineer starting from an ~81% baseline is already more exposed than a high-volume manufacturing hire near 91%. Combine that with scarcity and a long acceptance-to-start gap and you have the profile most likely to leak.

The procedure

Run this end to end for every offer, and re-run the scoring steps after each touchpoint, because fall-off risk rises across the acceptance-to-start gap and a stale score misses late competing offers.

The securing action ladder

The delivery channel of the offer is itself a lever, and a cheap one. Candidates who receive a personal call from the hiring manager before the written offer accept at 79% versus 61% for a letter only. That 18-point swing is free, so make the hiring-manager call your default action for any medium-or-higher score.

LOW  -> Standard monthly keep-warm to start date. [recruiter]
MEDIUM -> Hiring-manager personal call before/around written offer + accelerate paperwork + confirm start date in writing. [recruiter + hiring manager]
HIGH -> Accelerate start date, close the economic gap if pay is the driver, escalate to talent lead, AND keep runner-up warm. [talent lead]

Keep-warm cadence

Keep-warm begins immediately after acceptance, because the period between acceptance and start is when uncertainty develops most, especially if communication is poor. Default to a once-a-month schedule, ramp up in the final one to two months before onboarding, and scale down when a candidate is at risk of communication burnout. A named contact throughout beats a rotating cast.

How this goes wrong

The failure modes below are where a fall-off read produces the wrong action. Each has a false positive and a local check you can run in seconds.

• Treating short tenure as a verdict. Rejecting or over-securing a serial mover who is genuinely committed. Check against Cornell's 0.22 correlation - tenure alone explains almost none of the variance, so require a behavioural signal before you act on it.
• Trusting a verbal acceptance as secured. Closing the backup after a "yes" that is not signed. Check: is there a countersigned offer and a confirmed start date? If not, the offer is not secured.
• Reading fast, polite replies as commitment. A candidate can be responsive and still be shopping, since 47% stay open after accepting. Check: ask directly about competing timelines rather than inferring commitment from tone.
• Assuming a strong offer beats a counteroffer. Roughly 50% of resigners get countered; if you did not model the counter economics, your "secured" high performer is exposed. Check: did the candidate cite pay as the primary motive? If so, counter risk is high.
• Citing "80% leave within six months" as fact. That figure is unsupported by systematic data. Check: use the traceable CEB/HBR 12-month, 50% figure and label the rest as folklore.
• Over-communicating in keep-warm. Weekly emails read as reassurance to the recruiter but as desperation to the candidate, eroding boundaries. Check: default monthly, ramp only in the final one to two months.
• Scoring once at offer and never re-scoring. Fall-off risk rises across the acceptance-to-start gap, so a stale score misses late competing offers. Check: re-score after each touchpoint.

Verify before you call an offer secured

Run this checklist before you close the backup or tell a hiring manager the seat is filled. It catches the failures above at the point they cost you the most.

Keeping the read current

The dimensions in this framework are stable, but the numbers around them move, so re-anchor the read rather than trusting a value you memorised. Your own offer-to-acceptance rate is the first thing to track: compute it as offers accepted divided by offers extended times 100, watch it against the 69.3% NACE cross-industry average, and let a falling number tell you your securing actions are miscalibrated before individual reneges do.

Two mechanisms are worth re-checking on a schedule. Skill scarcity shifts as talent pools grow, so re-run the profile-count comparison for your critical roles each quarter rather than assuming last year's thin slice is still thin. Sourcing a warm backup pool of runner-up candidates keeps the highest-risk offers from becoming reopened searches, and describing that pool in plain English - runner-up-quality engineers in your metro who are open right now - is faster than rebuilding a shortlist under pressure. The read is only as good as its freshest input, so treat the score as a living value you refresh at every touchpoint, not a stamp you apply once at offer.