Pick One Clock, Write Down Its Start, Stop Benchmarking Outward

Which clock does each metric use?

Time to fill spans the requisition being opened to the offer being accepted. Time to hire spans one candidate's entry into the process to that same acceptance. They end in the same place, which is why the names get swapped, and they start in different places, which is why they answer different questions. One describes your staffing pipeline. The other describes your process.

SHRM's benchmarking report is explicit about this. It measures time to fill as the number of days from the job requisition being opened to the offer being accepted, calculated in calendar days including weekends and holidays ¹. Two things follow immediately: internal approval delay is inside the number, and everything after acceptance, including a notice period and a start date, is outside it.

So the arithmetic difference between the two metrics covers two legs. The first is the intake and approval overhead, the days between a role being agreed and a candidate being able to apply for it. The second is the wait from a live posting until the candidate in question actually arrived. Neither is owned by anybody in particular, and both sit outside any benchmark whose clock starts at the application.

Pick based on the question being asked. A board asking why a team is understaffed is asking a time-to-fill question, and the approval leg belongs in the answer. A recruiting lead asking whether the process is efficient is asking a time-to-hire question, and including approval delay there charges the recruiting team for a decision made somewhere else.

Why are published day benchmarks not comparable?

Because each figure runs a different clock over a different population, and nothing published lets you reconcile them. A benchmark in days needs its start point, its end point, whether the days are calendar or working, and whose hires sat in the sample. Most quotations of one carry none of those four, which is what makes them so easy to repeat.

Three published figures, and what each actually counts:

• SHRM's 2026 recruiting benchmarking brief, built on data from over 4,600 organizations, reports a median time to fill of 39 calendar days for nonexecutive positions ². That is a median across organizations answering a survey, so an employer making two hires a year weighs exactly as much as one making two thousand ².
• Ashby's median Time to First Fill is 56 days for business roles and 76 for technical roles, counted from a job being opened to the first hire being made ³. Stopping at the first hire makes evergreen and multi-hire requisitions look faster than they are, and Ashby says the timeline is hard to measure consistently because a hiring effort can be paused, restarted, or need a new job description ³.
• The DHI-DFH vacancy duration measure, built on the federal job openings and labor turnover survey, put mean US vacancy duration at 28.3 working days in June 2016, against 59.5 working days at establishments with more than 5,000 employees ⁴. Working days rather than calendar days, a national establishment-level aggregate, and a series that has since been discontinued ⁴.

Hold those beside each other and the trap is obvious. The distance between them is not a disagreement about how long hiring takes: the clocks differ and the samples differ at the same time, and no published source apportions the gap between those two causes ²³. Quoting any of them as your target imports somebody else's definition along with their employers. Which stage-level metrics survived the change in application volume is the same problem one layer down.

Which leg actually grew?

A total cannot tell you, and it is the only number most reports contain. Elapsed time is a sum of legs that move independently: approval, posting to first qualified candidate, screening, interviewing, and decision through to acceptance. Those legs can swap sizes across a year while the total holds steady, so a flat line is evidence of nothing until each leg is recorded on its own.

The scheduled legs are smaller than people expect. In SHRM's 2021 benchmarking data, a different survey year and a different sample from the 39-day figure above, the median organization 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, inside a median time to fill of 44 days for nonexecutive roles across 840 organizations, with a 25th percentile of 28 days and a 75th of 73 ¹. Each of those four medians is a week or less, against a median total of 44 days.

Interviewing, the part everyone argues about shortening, is a small share of the calendar. The waiting is in approval, in the gap before a qualified candidate arrives, and in the pauses between scheduled events that nobody logs as a stage at all.

The arithmetic is worth doing on paper once. A screening leg that halved and a decision leg that doubled produce exactly the same headline as a year in which nothing moved. Which of your legs did what is a question for your own data rather than an assumption in either direction: where the time goes when every stage got faster is one worked case of a total moving for reasons nobody expected.

Take twenty hires and mark the timestamps

Pull the last twenty hires and mark six moments for each: requisition approved, posting live, first qualified candidate, decision made, offer accepted, start date. Then name the leg that grew. Twenty is enough to expose a leg that doubled and few enough that one unusual search will not dominate the picture, and the exercise takes a morning of somebody's time.

Then make those six moments fields in the applicant tracking system, so nobody has to rebuild them by hand. A number rebuilt from email threads changes every time somebody rebuilds it, which is how two people end up quoting different figures from the same quarter and neither of them is lying. Fields also survive the person who ran the analysis leaving.

Report one clock with its definition at the top, and print the number of hires beside the median. Split executive from nonexecutive before anyone reads a trend: in the same benchmarking data, executive roles ran a median of 60 days across 666 organizations against 44 for nonexecutive ones ¹, so a quarter weighted toward senior searches produces a longer number with nothing changed about the process.

And resist the urge to add a second headline metric. One clock, five legs, a definition line and a hire count is a report anybody can argue with productively. Two totals with different clocks in the same deck guarantee that the meeting is about which number is right rather than about which leg to fix. If the leg that grew is the one before a qualified candidate arrives, what happens when a posting hits its application cap in two days is where that investigation starts.

See a sample report >