Pipeline
The Longest Gap Between Stages Is the One a Candidate Feels
A hiring process is moving fast enough when no single gap between stages runs longer than about a week and every active candidate holds a stated next-step date. Published medians for a whole requisition run from 39 days to 76, depending on who counted and what their clock measured, so no total works as a target. What a candidate experiences is one silent stretch, owned by an interviewer's calendar or by a decision nobody scheduled. Put a service level on each gap and publish the elapsed time you expect.
The takeFaster is not uniformly fairer, and the benchmark posts never price that. Compressing a loop into a single week shifts the burden onto candidates with the least control over their own schedule: shift workers, people with caregiving duties, anyone who cannot take four weekday hours at short notice. Speed presented as a pure good hides who paid for it. My read is that predictability buys everything speed was supposed to buy, and it buys it without picking whose calendar has to be flexible.
Where Olive fits
Open a role and see what the work shows
Olive's session runs async on the candidate's own clock, 50 to 70 minutes, so it adds nothing to anyone's scheduling queue. What comes back is six findings written by a reviewer, each carrying the excerpt it rests on, and the candidate is given the same document.
Rank your shortlistHow fast does a hiring process need to move?
Fast enough that no gap between stages runs past about a week. That is a different unit from a total, and it is the unit a candidate actually experiences. Nobody sits through a forty-day process as forty days. They sit through the nine days in the middle when nothing arrived and no date had been promised.
SHRM's benchmarking data is useful mainly for showing where the days are not. Median time-to-fill for nonexecutive roles was 44 days, with a 25th percentile of 28 days, a 75th of 73 and a mean of 54, and inside that the median organisation spent 5 days screening applicants, 7 days conducting interviews, 4 days making a final decision and extending an offer, and 2 days from offer to acceptance 1. Those are four separate medians, so they do not add up to the 44. What they show is the ordering: every hands-on activity is measured in single-digit days, while the requisition itself stays open for weeks.
Read the caveats before quoting any of it. SHRM's clock starts when the requisition opens and stops when the offer is accepted, so it includes internal approval delay and excludes notice periods. It is 2021 data from a membership association's own organisations. And the median and the mean sit ten days apart, with quartiles at 28 and 73, so no single figure describes that distribution. What survives all of it is the shape: interviewing occupies a small share of the elapsed calendar, and locating the rest of it is your own measurement job.
Set a service level on each gap, not on the loop
Write a target against every gap and report the worst one weekly, named by stage. Two commitments close the intervals that go quiet: a substantive reply within two business days, and the next step booked the moment a stage passes, ahead of the debrief. Neither asks anyone to decide faster. Both give the candidate a date to hold while the deciding happens.
The gaps worth instrumenting are the ones your applicant-tracking system can already see:
1. Application to first human response. The one that has inverted. Across one applicant-tracking vendor's mostly technology customers, the average recruiter now processes 291 applications per hire against roughly 100 in early 2021, so the queue forms on your side 3. 2. Screen to first interview. Almost entirely a scheduling artifact, and the one most improved by sending three slots the moment the screen ends. 3. Last interview to decision. The one that goes quiet while a debrief waits on a single calendar. 4. Decision to offer delivered. Approval work, invisible to everyone except the candidate.
Campus hiring gives a sense of what a well-instrumented version looks like: NACE's 2023 benchmarking survey put the average time between a candidate's first interview and an offer or a rejection at 25.1 days, with students taking about 11 days to accept and reneging on 8% of accepted offers 4. That is US college and entry-level recruiting only, self-reported by employers who pay to belong to a campus-recruiting association, with 269 of 889 members responding. It does not transfer to other hiring. What it shows is an employer measuring the interval a candidate sits in.
If your own gaps have grown while every stage got tools that were meant to shorten them, the diagnosis is worth doing properly: where the time actually went after AI entered every stage and why applications per opening tripled are the two halves of that answer.
Which time-to-hire benchmark are you being measured against?
Whichever one your board read most recently, and they do not measure the same thing. SHRM's 2026 brief reports a median time-to-fill of 39 calendar days for nonexecutive positions, drawn from over 4,600 organisations 2. One large applicant-tracking vendor reports a median time to first fill of 56 days for business roles and 76 for technical ones 3. Both numbers are real. Neither of them is yours.
They differ for two reasons at once, and this is the part that gets flattened. The clocks differ: time to first fill stops at the first hire on a requisition, which makes evergreen and multi-hire roles look faster, while SHRM's clock runs from requisition open to offer accepted. The populations differ too: a US human-resources membership association's respondents on one side, one vendor's venture-backed technology customers on the other. Nothing published lets you apportion the gap between those two causes, so quoting them as rival estimates of one quantity is simply wrong. SHRM's figure is also an organisation-level median, which means an employer making two hires a year weighs exactly as much as one making two thousand.
Match the definition or drop the comparison. Matching it costs more than measuring your own gaps does, and the gaps are the part you can act on. Measure the same clock every quarter, report the whole distribution with its quartiles, and treat any external figure as background. A req that closes suspiciously fast deserves the same scrutiny as one that drags: a posting that hit its application cap and closed in two days is a speed problem too, in the other direction.
Never compress the write-up
The one part of a loop that should never be squeezed is the record. An offer extended before the evidence is written down is fast and unaccountable, and six months later, when someone asks why, the record is what answers them. Have scorecards written before the debrief opens, and let the calendar absorb the day that costs.
Compression has a distributional cost as well, and no benchmark post prices it. A loop demanding four weekday hours inside five days is easiest for people who are not currently working and hardest for people who need to arrange cover, childcare or a shift swap. An expected elapsed time published in the posting, a choice of slots including outside core hours, and a stated next-step date give a candidate what speed was supposed to give them without moving the cost onto whoever has the least schedule control. The other half of that budget is measured in hours, which is the whole loop measured in candidate hours.
One honest limit sits over all of this. Acceptance rates move with the labour market as much as with anything a recruiter does, so a quarter where offers land better is no proof the loop improved. Measure what you control, which is the gaps and the record.
On Monday, instrument the three gaps your system can already see, name the longest one by stage in the weekly hiring note, and give every active candidate a date instead of a promise to be in touch. The first week of that report is usually uncomfortable and always specific, which is the combination that gets a calendar unblocked.
Common questions
What is a reasonable total elapsed time for a normal role?
Published medians run from 39 days to 76, on clocks that all start when a requisition opens rather than when a candidate first hears from you, so the honest answer is a range and a caveat. Treat it as background. A six-week loop with no gap longer than five days feels responsive to a candidate, while a three-week loop containing one eleven-day silence does not. The figure worth publishing is the longest gap, because that is the one a candidate can tell whether you kept.
Which single metric should a talent dashboard carry?
The longest inter-stage gap in the last week, named by stage. It points at a specific calendar rather than at the function, which is what makes it actionable. Keep total time-to-fill alongside it, reported as a distribution with quartiles, since these distributions are heavily skewed: SHRM's nonexecutive quartiles sit at 28 and 73 days around a 44-day median with a 54-day mean. Add offer acceptance if you want, and read it as a market signal about the quarter.
Does a faster process win more candidates?
It removes a reason to leave, which is not the same thing as winning one. What a fast loop actually delivers is fewer gaps and a candidate who always knows what happens next, and both of those are available without the compression. Compensation, the role and the manager are all still in that decision. The honest framing for a leadership team is that a slow, silent loop loses people you had already earned, so fixing it protects a pipeline you already paid for.
Is it fair to compress a loop into a single week?
Only if you offer alternatives alongside it. A one-week loop suits candidates between roles and penalises anyone currently working, caring for someone, or on a fixed shift, and that filtering never appears in a funnel report because those candidates stop replying rather than complain. If speed genuinely matters for a req, publish the compressed timeline, offer slots outside core hours, and let candidates request a longer schedule without treating the request as a signal about their interest.
Where does approval delay belong in this?
Inside the measurement, because the candidate experiences it. Requisition approval, offer approval and compensation sign-off routinely add days that nobody in recruiting controls and everybody in recruiting gets blamed for. Report them as their own gaps with the approver named, and the conversation moves from a vague complaint about hiring being slow to a specific one about a step that takes four days. Delay tends to shrink once it has an owner attached to it.
References
- 1. SHRM Benchmarking: Talent Access (Selection Criteria, Overall) shrm.org Supports the time-to-fill median, quartiles and mean for nonexecutive roles and the per-activity day counts showing that interviewing is a small share of elapsed time.
- 2. 2026 Recruiting Executives Benchmarking: Attracting Critical Talent shrm.org Supports the 39-calendar-day median time-to-fill and its sample of over 4,600 organisations, cited as a median across organisations rather than across hires.
- 3. Recruiter Productivity | 2026 Talent Trends Report ashbyhq.com Supports the median time to first fill for business and technical roles, the definitional difference that stops it being compared with a time-to-fill median, and the 291-applications-per-hire figure against roughly 100 in early 2021.
- 4. 2023 Recruiting Benchmarks Report Executive Summary naceweb.org Supports the interview-to-decision interval, days to accept and renege rate in campus hiring, quoted with the survey's sample and response rate.
4 sources, numbered by first appearance. How Olive sources claims
General guidance for hiring teams. What works at one company and one volume may not transfer to yours.
Olive assesses how a person works with AI. It does not detect AI-written documents, and it never produces a score, a ranking, or a match percentage for a person. Candidates read the same report the employer reads.