They Signed, Then Vanished: The Offer-to-Start Dead Zone

The offer-to-start-date dead zone is the stretch between a candidate accepting your offer and their first day, typically two weeks to three months. It is where most new-hire ghosting happens, and because nearly every applicant tracking system treats “offer accepted” as a terminal state, it is the one pipeline stage with no owner, no service-level agreement, and no data.

If you have had a signed candidate stop replying, you probably read it as a personal failure. It is not. It is a schema failure. The moment the offer comes back signed, your system of record marks the requisition closed, the runner-up gets a rejection email, the hiring manager updates the roadmap, and a human being you have never worked with disappears into someone else’s notice period for six weeks with nobody on your side owning them. Nothing in your tooling is watching. Then, three days before the start date, the Slack message arrives: “I’m so sorry, but…”

Why candidates accept an offer and then never show up

Most of them did not plan to. They accepted honestly, then spent weeks in a window where your competition kept selling and you went quiet.

Gartner’s research on candidates who backed out after accepting (n≈3,500, surveyed May 2022 to May 2023, reported via SHRM) found that nearly 90% cited at least one mismatch in the employee value proposition. The reasons were not primarily financial:

Money came third. Flexibility beat it, and flexibility is exactly what a candidate keeps learning about during the notice period: from friends, from Glassdoor, from the recruiter who is still calling.

The same research found that among roughly 2,000 people who had recently started a new role, 47% remained open to other opportunities and 42% believed they could find a better role by continuing to search. Jamie Kohn, senior research director at Gartner HR, describes candidates keeping “one foot in the job market” after saying yes. Acceptance is not an exit from the market. It is a pause.

Then there is the counteroffer mechanic, which is structurally invisible to you. Your candidate resigns. Their current employer, now facing a real backfill cost, comes back with money, a title, or a team change, and that conversation happens entirely inside a company you have no contact with, on a timeline you cannot see. You find out through silence.

LiveCareer’s survey of currently employed U.S. adults (Pollfish, n=1,008, fielded 12 February 2025) found one in four job seekers have ghosted an employer, with accepting another job offer the single most common reason. Not rudeness. Not chaos. A better option arrived in a window where you had stopped competing.

How often do candidates renege on an accepted offer?

Somewhere between a third and half of candidates, depending on which year you ask. The rate moves with the labour market, and it moves fast.

Gartner’s quarterly candidate research is the best longitudinal series publicly available:

Sources: Gartner via SHRM and Gartner HR Research, 1Q25.

Read that last row carefully, because most coverage of this topic gets it wrong. Reneging did not rise steadily to an all-time high. It fell 13 points between 1Q24 and 1Q25. Caroline Ogawa, director of research at Gartner HR, attributes the drop directly to the market: “We’re seeing signs of a softening labour market, with fewer candidates getting multiple offers or backing out after accepting a job offer.”

That reversal is the most useful fact in this article, and not for the reason you would expect. Reneging is not a character trend among candidates. It is a measure of who holds the bargaining power. More offers on the table, more reneges; fewer offers, fewer. So it will come back, on a schedule set by the hiring market rather than by anything you do.

Here is the operational problem with a metric that swings 13 points in twelve months: nobody builds tooling for it. When the market softens, the problem quietly disappears and the fix drops off the roadmap. When it tightens, every company rediscovers it at once, in the worst possible way, one no-show at a time. The teams that survive the next tightening are the ones that instrumented the window while it was cheap.

The dead zone: what actually happens between signature and day one

The window is long, it is where ghosting concentrates, and its length is set by your candidate’s current employer, not by you.

Click Boarding, a preboarding software vendor, commissioned a OnePoll survey of 2,000 U.S. adults who changed jobs in the previous two years. 45% of those who ghosted did so between receiving the offer and their first day, which the survey describes as more than twice the rate of any other stage. The same survey found 53% admitted ghosting a new employer at some point, and that 61% waited at least two to three weeks to start, with 29% waiting more than a month.

Treat those numbers as directional. The vendor sells preboarding software, no methodology page is published, and the same article carries an unsourced cost table we would not cite. But the wait-time distribution is the part that matters, because it defines the window you are failing to manage.

That window varies enormously by jurisdiction, which is why “stay in touch for a couple of weeks” is bad advice in half the world:

• United States: at-will employment, two weeks is a customary courtesy rather than a legal requirement.
• United Kingdom and much of the EU: statutory and contractual notice commonly runs one to three months. Three-month notice is standard in mid-to-senior enterprise technology roles.

A U.S. startup hiring a senior engineer out of a European enterprise faces a twelve-week gap between signature and day one, longer than the interview process that produced the hire. That is long enough for their team to counteroffer, for a recruiter to place them elsewhere, and for your entire pitch to fade into a PDF they signed once.

Nothing in that window is unusual. What is unusual is having a system that notices.

Why your ATS can’t see any of this

Because “offer accepted” is modelled as a terminal state, not as a handoff into another stage that still has work in it.

This is a category-level truth, not a competitor jab. Greenhouse, Ashby, and Lever all effectively end their pipeline model at offer acceptance, leaving the acceptance-to-onboarding handoff to a separate onboarding tool, an HRIS, or a recruiter’s memory. The point is structural: the applicant tracking system stops tracking when the applicant stops being an applicant. They are not an employee yet, so no system owns them.

Kit is in the same position, and we are not going to pretend otherwise. Run grep start_date db/schema.rb against our own codebase and you get nothing. The single most important date in the entire hiring process is not a field. Not in ours, not in most of the products we compete with. Everything downstream of that absence follows: you cannot compute days-to-start, you cannot flag a start date that is drifting, you cannot report on how many accepted offers turned into people.

So the offer-to-start window is not managed badly because recruiters are lazy. It is managed badly because the system of record marks the record closed, and everything after that runs on a coordinator remembering to send a calendar invite.

Offer acceptance rate is not start rate

Offer acceptance rate measures the percentage of extended offers that candidates accept. Start rate measures the percentage of accepted offers that become humans in seats. The gap between them is invisible in most ATS reporting because the record closes at acceptance.

Almost every talent team reports the first number to their exec team. Almost none can compute the second one, which means the number on the slide is systematically flattering.

Put money on it. SHRM’s 2025 benchmarking research puts average cost per hire at $5,475 for non-executive roles and $35,879 for executive roles, with time-to-fill running roughly a month and a half. A no-show does not cost you $5,475. It costs you:

1. The original $5,475, fully sunk, with no hire attached.
2. A second $5,475 to run the search again.
3. Another 45 days of time-to-fill, starting from an empty shortlist rather than from your finalists.
4. The quarter of roadmap the hiring manager planned around a start date that never happened.

That is roughly $11,000 and three months for one non-executive role, and it never appears in a single hiring report, because the report ends at the acceptance you already counted as a win. For a startup making eight hires a year, one no-show is a 12.5% failure rate on the entire hiring function, recorded internally as 100% offer acceptance.

You do not need a new philosophy to fix this. You need a stage.

What a pre-boarding stage looks like: owner, SLA, cadence

A pre-boarding stage is a real pipeline stage that runs from offer acceptance to day one, with a named owner, a communication SLA, and a due date. It is not a flag on a closed record.

The precedent already exists inside most hiring processes: the reference check. Reference checks are work that happens outside your building, on someone else’s clock, with a deadline you do not control and a completion threshold you define. Nobody models a reference check as a boolean on a closed requisition; it gets a stage, an owner, and a nudge. Pre-boarding is exactly the same shape and roughly ten times more expensive to get wrong.

Give the stage an owner

Not “the recruiter, probably.” A named person, recorded on the record, who is accountable for the candidate showing up. In most startups this defaults to whoever wears the ops hat, which means it defaults to nobody. Write the name down. Ownership that is not in the system is not ownership.

Set a communication SLA

Decide the maximum acceptable silence and enforce it. For a two-week U.S. notice period, weekly contact is reasonable. For a twelve-week European one, go weekly for the first fortnight, then fortnightly, then daily in the final week, with a concrete artifact each time: the laptop order, the first-week schedule, an intro to their future teammate. The failure mode is not too little warmth. It is nothing for five weeks and then a form on the Friday before.

If you have already built response-time discipline for the top of your funnel, this is the same machinery pointed at the other end. Our guide to hiring SLAs and employer ghosting covers how to set thresholds that people actually meet.

Escalate when a signed candidate goes quiet

Silence after acceptance is the highest-signal event in your entire pipeline, and almost no team treats it as an alert. A signed candidate who does not reply to two consecutive touchpoints in the final fortnight before a start date is telling you something. Escalate to the hiring manager, not to a nurture email. The conversation you want is a phone call from the person they will actually work for, and you want it while the counteroffer is still being weighed, not after it has been accepted.

There is one prerequisite most teams miss: the channel has to still be open. If your candidate portal login expires a week after acceptance, you have architecturally guaranteed that the most important conversation in the process happens in a recruiter’s inbox, unlogged and unmeasured. This is a real constraint in our own product, and we say more about it below.

Capture why they left, or you’ll never fix it

Record a structured reason on every renege, chosen from a fixed list, at the moment it happens. Free-text notes in a closed record are not data.

Six values cover almost everything:

• Counteroffer from their current employer
• Competing offer from another company
• Compensation on your side, revisited after signature
• Relocation or logistics collapse
• Background check or paperwork delay that outlasted their patience
• Went silent with no stated reason

The value of the taxonomy is that each entry points at a different fix. A cluster of counteroffers means your offer conversation is not addressing why they are leaving, only what they are getting. A cluster of competing offers means your dead zone is too quiet. A cluster of paperwork delays means your ops process is the thing losing you engineers, which is both the most embarrassing cause and the easiest to fix.

With that field in place, you can finally compute the number that matters: accepted offers divided into people who actually started, reported next to offer acceptance rate, on the same slide. The first time you see both numbers together is usually the last time you trust the first one alone.

When a renege lands: reopen from your silver medalists

Start from your finalists, not from a fresh job posting. Your ATS already holds a fully evaluated shortlist that you rejected on the day the offer was accepted.

This is the second-order cost nobody budgets for. The runner-up was told no six weeks ago, and by the time the renege surfaces, often days before the start date, they have usually taken another role. The requisition does not restart from the shortlist. It restarts from zero, on a 45-day clock, with a hiring manager who has already told the team the seat is filled.

Two things reduce the damage. First, do not send a hard rejection to your top three the moment an offer is accepted. Tell them the truth: the role is provisionally filled, you were impressed, you would like to stay in touch. Second, mine the pool properly when the seat reopens. A candidate who reached your final round nine months ago is a warmer, faster, cheaper start than any new applicant, and most teams never look. Our piece on talent rediscovery and silver medalists covers how to work an existing database instead of reposting.

The uncomfortable version of this: your best defence against a renege is not the nurture email. It is having a second candidate who has not been sent away.

How Kit thinks about the offer-to-start window

Kit is an applicant tracking system built for startups, and we are going to be specific about which parts of this argument are shipped and which are a position we hold.

What exists today. Stage types in Kit are first-class and configurable, including a dedicated offer stage. Offers are real records with acceptance, decline, and expiry timestamps, not a status field, and candidates accept in the portal rather than in an email thread, so acceptance is a timestamped event instead of a message someone has to notice. The reference-check stage already models an owner, a deadline in days, and a completion threshold, which is the proof that the pattern in this article works. Kit also ships anti-stall nudges with configurable escalation, currently pointed at employer-side silence in the middle of the funnel.

What does not exist yet, in Kit or in the products we compete with. There is no start_date field in our schema. Our hiring analytics stop at applications and conversion rate, so start rate is not computable today. There is no structured renege reason on an offer record: we store declined_at without a cause. And our candidate magic links expire after seven days. Against a notice period where 29% of people wait more than a month, that closes the authenticated channel for most of the dead zone by construction. That last one is the sharpest gap, because the passwordless portal is exactly the right place for start-date logistics, paperwork, and questions to live, and right now it shuts too early.

We are writing this while the market is soft and the renege rate has fallen to 35%, which is precisely the wrong time for anyone to care and precisely the right time to build. The argument stands on its own: the offer-to-start window is the only stage in hiring with real money attached and no instrumentation, and it is unmeasured everywhere. If you want to see the candidate-side version of the same blind spot earlier in the funnel, start with why candidates ghost you.

If you want a pipeline where stages, owners, and deadlines are configurable rather than assumed, try Kit free. And if you build the pre-boarding stage before we ship the field, tell us what you learned. We would rather ship the right thing late than the obvious thing wrong.