It's Tuesday standup, and the same three tickets roll over for the fourth sprint in a row. The recruiter's update hasn't changed since April: the pipeline looks good, some strong candidates in late stages. This is the third month the pipeline has looked good, and the roadmap has stopped believing it. If it feels like you can't hire engineers fast enough, the numbers say you're not imagining it.
Key takeaways
Hiring a software engineer takes a median 75 days for technical roles, versus 60 for business roles. Senior searches run 71 days to first fill, and the interview loop alone takes a median 17.9 days [C1][C2][C3]. Candidate notice periods stack on top of all of it.
Here's the number underneath that standup: technical roles take a median 75 days to first fill, versus 60 for business roles [C1]. And the clock stretches with seniority: 52 days for junior roles, 63 for mid-level, 71 for the senior engineer you actually posted the req for [C2]. Even once you've found the person, the interview loop alone runs a median 17.9 days from first interview to last [C3].
Those figures come from Ashby, which means they're drawn from Ashby's own customer base: mostly venture-backed, tech-forward companies. Which is a caveat and also, probably, a description of your market.
So the 75 days isn't a horror story. It's the median. Half of searches run longer. And while the search runs, the open seat bills you in three currencies nobody invoices: the roadmap that doesn't ship, the team that absorbs the load, and the hire you make under pressure.
Let's read each bill.
No invoice arrives for the feature that slipped a quarter. That's what makes this bill dangerous: it never hits a budget line, so nobody defends against it.
You can put arithmetic on it, though. The standard cost-of-vacancy method — published in Stack Overflow's employer-branding resources back in 2022 — is annual salary ÷ 260 working days × days open [C4]. Run it on a $180K senior: $180,000 ÷ 260 ≈ $692 per working day. Ashby's 75-day median is roughly 54 working days, so that's about $37K in salary-equivalent output that never existed. Your number will differ — run your own. And to be clear about what this is: it's your arithmetic, not a research finding. There is no rigorous study that hands you a cost-of-vacancy figure; there's a formula and your own inputs.
Some models go further and apply role-based impact multipliers: roughly 2× for a revenue-generating engineer, up to 2.5× for a senior IC, on the logic that an engineer's output is worth more than their salary [C5], though that comes from a calculator site's model, not peer-reviewed anything. And one offshore-staffing vendor — selling the fix, mind you — pegs a four-month senior search at $60,000–$100,000 in opportunity cost [C6].
The honest summary: the exact figure is unknowable. The direction and the scale are not. An open senior seat is burning something in the tens of thousands per quarter, and it's burning it out of the roadmap, which is the one place you can't backfill with budget.
The missing person's tickets don't vanish. LeadDev's 2025 survey of 617 engineering-org respondents put it plainly, writing about layoff survivors: when colleagues are laid off, their work gets "reallocated to those who remain" [C10]. An unfilled seat runs the same mechanism from the other direction. You've watched this happen. The two seniors who sat nearest the gap quietly picked it up, because they're the kind of people who do.
The same survey shows what that absorption costs: 38% reported working longer hours while only 7% worked less, 65% had their responsibilities expand, and 22% rated themselves at critical burnout, with another 24% at moderate [C10]. Harness's State of Developer Experience survey (500 engineering leaders and practitioners, 2024) found 23% of developers working overtime ten or more days a month [C9].
Now the part that should genuinely worry you. Burnout doesn't just degrade output — it converts to attrition. In that same Harness survey, 52% of developers named burnout as a primary reason their peers left their jobs [C8]. (Harness sells developer-experience tooling, so read with one eyebrow raised — but the survey went wide, and the finding matches what every engineering leader has seen up close.)
Close the loop on what that means: an unfilled seat overloads the people covering for it, overload becomes burnout, and burnout becomes a resignation letter from someone you couldn't afford to lose. Which opens a second seat. Which starts a second 75-day clock [C1]. Absorption doesn't save the roadmap. It doubles the vacancy risk.
Month three. The roadmap is bleeding, your two best people are tired, and the next candidate in the loop is a 7-out-of-10. Six months ago you'd have passed without a second thought. Today there's a voice in the debrief saying we can coach the gaps.
That voice is the vacancy talking. Pressure lowers the bar — not because you're weak, but because the cost of the open seat is now visible every single day and the cost of a mishire is hypothetical.
You've heard the folklore percentages about what a bad hire costs. They're widely cited and rarely sourced, so skip them and count the thing you can actually count: a mishire restarts the whole clock. Another req. Another 75-day median search [C1]. Another 2.5-week interview loop [C3]. Another 3–6 months of ramp before the replacement reaches full productivity [C11]. Made in month three, a compromise hire doesn't resolve the vacancy; it extends it deep into next year, with a severance conversation in the middle.
Hiring while your roadmap stalls is negotiating with a gun to your own head. The fix isn't more discipline in the debrief. It's removing the gun.
The instinct is to push the search harder. More sourcers, more outreach, tighter SLAs on feedback. Some of that helps at the margins. But look at where the time actually lives.
Most of the clock isn't yours. Candidate supply, competing offers, notice periods: market conditions you don't control. Even the part you do control has a floor: once you've found the right person, the loop itself still takes a median 17.9 days [C3], and no amount of urgency compresses a candidate's four-week notice period.
"Fix the funnel" is the other standard answer, and it's correct — and slow. A full employer-branding program runs 8–10 months, sometimes longer [C12]. That's longer than the vacancy it's meant to shorten. Do it anyway; it fixes the next search. It does nothing for this one.
And here's the part that reframes the whole problem: day 75 isn't the finish line. Industry-average ramp to full productivity is 3–6 months, with strong onboarding compressing it to 8–12 weeks [C11]. So the real gap between "req opened" and "productive engineer" is roughly 6–9 months once notice periods stack on. That's not a hiring delay. That's a planning horizon.
Which points at the move that actually is yours to make: stop letting the roadmap's fate ride on the req. Decouple them. Keep the search running at whatever speed the market allows, and put bridge capacity on the roadmap now — both, not either. We've laid out the full bridge-capacity math separately if you want to pressure-test it.
Before this sounds too clean: a bridge engineer pays a context tax too. Ramp applies to everyone [C11]: nobody ships like a two-year veteran on day one, including whoever you bring in to cover the gap.
Task-scoped work narrows that tax. A well-scoped ticket — clear inputs, defined done, your review gate — needs far less organizational context than an embedded seat that has to absorb your architecture history and your team's unwritten rules. Narrower, not zero.
And a bridge doesn't replace the hire. Core-IP ownership, long-horizon architecture calls, the person who carries the system in their head — that still belongs with someone permanent, and you should keep searching for them. (When to hire versus when to bridge is its own decision; our staffing-models guide walks through it.)
So the real choice was never context-tax versus no context-tax. It's whether you start paying ramp this week, or after day 75.
Concretely, here's what decoupling the roadmap from the req looks like with DevOD: a same-day match, first task shipped within five business days, then a 3-day task cycle with daily async updates and a task-by-task approval gate: you approve each piece of work before the next begins. It runs month-to-month and you can cancel any time, so when your hire finally signs, the bridge comes down. No lock-in dressed up as a contract. A single stream runs $3,495/mo, and it's one lane of the broader on-demand product team model.
Don't take the pitch on faith — take it on output. Hand over one real ticket from the pile that's been rolling over since April and judge what comes back: it starts with a 15-minute fit call and a one-task Proof of Quality before any subscription.
Keep running the search. Interview well, hold the bar, hire the 9. Just stop making the roadmap wait for it.
Technical roles take a median 75 days to first fill, versus 60 days for business roles [C1]. Seniority stretches the clock: 52 days for junior roles, 63 for mid-level, 71 for senior engineers [C2]. The interview loop alone runs a median 17.9 days from first interview to last [C3].
Divide annual salary by 260 working days, then multiply by days open [C4]. A $180,000 senior costs about $692 per working day, or roughly $37K over a 75-day median search (~54 working days). Some models add a roughly 2×–2.5× role-based multiplier for revenue-generating engineers [C5].
The work gets reallocated to those who remain [C10]. In LeadDev's 2025 survey, 38% reported working longer hours, 65% had responsibilities expand, and 22% rated themselves at critical burnout [C10]. In Harness's survey, 52% of developers named burnout a primary reason peers left their jobs [C8].
Run both: keep the search at whatever speed the market allows and put bridge capacity on the roadmap now. Task-scoped work narrows the context tax versus an embedded seat, but core-IP ownership and long-horizon architecture calls should still go to the permanent hire you keep searching for.
Industry-average ramp to full productivity is 3–6 months, with strong onboarding compressing it to 8–12 weeks [C11]. Stacked on a 75-day median search, the real gap between "req opened" and "productive engineer" is roughly 6–9 months once notice periods stack on [C1][C11].
