A declined offer costs more than any other failure in the funnel. Every hour of screening, interviewing and debriefing is spent, the role is still open, and you usually restart at the top.
The instinct is to fix the offer. Almost every cause is upstream of it.
What a healthy rate looks like
Above roughly ninety per cent suggests you are only extending offers you are certain about — good, unless it means you are being slow and losing people before the offer stage. Eighty to ninety is a normal, healthy band. Below eighty, something systematic is wrong.
Segment before drawing conclusions. A blended rate across junior operations and senior engineering hides the segment that is actually failing.
1. Establish compensation expectations in the first conversation
The most common cause of a declined offer is a mismatch that was discoverable in week one and went unexamined until week six. Publish the range in the posting, confirm it on the screening call, and record the candidate expectation on the record where the hiring manager can see it.
The conversation is uncomfortable once and saves a wasted process. If a candidate wants twenty per cent above your band, you both benefit from knowing in the first ten minutes.
2. Compress elapsed time
Acceptance rate falls with process length, for a mechanical reason: the longer your process runs, the more likely another one concludes first. Strong candidates are usually in more than one process.
The recoverable time is rarely in the interviews. It is in the queue before first review, the scheduling gap, and the wait between the last interview and the decision — see debriefs that produce decisions.
3. Find out what they are actually optimising for
Ask directly, mid-process: what would make you choose one offer over another? Candidates answer honestly when asked plainly, and the answer is frequently not money — it is scope, the manager, remote flexibility, or a specific technology.
Record it. An offer conversation that addresses the thing they told you mattered is a different conversation from one that leads with a number.
4. Let them meet the manager and the team
People accept jobs and leave managers. A candidate who has spent real time with the person they will report to has enough information to say yes with confidence. A candidate who met four interviewers and no future colleagues is deciding on a brochure.
5. Make the verbal offer a conversation, not a number
Call before sending paperwork. Cover why the panel wanted them specifically, the scope of the role, the compensation, and what the first three months look like. Then ask directly whether anything gives them pause.
The last question is the one that matters. Hesitation surfaced on a call is negotiable. Hesitation discovered when the deadline passes is not.
6. Set a deadline that is short and real
Three to five working days for most roles. Long deadlines are not generous — they are an invitation to use your offer as leverage elsewhere, and the leverage usually works.
If a candidate asks for more time, ask what they are waiting for. The answer tells you whether you are the first choice, and whether extending helps you.
7. Stay in contact between acceptance and start date
Acceptance is not the end. Where notice periods run to one, two or three months, that window is long enough for a counter-offer or a change of mind — and the risk is concentrated in exactly the markets where notice periods are longest.
Low-effort contact works: a message from the manager in week one, practical onboarding information, an invitation to a team event. Then treat pre-start attrition as a tracked number rather than an occasional surprise.
Learning from declines
Record a structured reason on every decline: compensation, competing offer, role scope, location, manager, timing, counter-offer, other. Free text will not aggregate.
After ten declines the pattern is usually unambiguous, and the fix is rarely what people assumed. Teams convinced they had a compensation problem often find they had a speed problem — and those need entirely different responses.
For tracking this alongside the rest of the funnel, see reporting and the wider metric set.