How Long to Leave a Job Posting Open Now

Set the Close by Volume, Not by Days

Pick the number before you publish. Two readers, four minutes an application, four hours each in the first week: that is roughly 120 applications, and 120 is your cap. When the cap is reached, close the posting or pause it. If the read produces nobody worth advancing, reopen and take the next tranche, with the rule adjusted and the change written down.

The reason to fix this in advance is that nobody chooses well at the moment of overload. A posting that has taken 600 applications by Thursday produces a decision made in fatigue, and the usual outcome is neither reading nor closing: the requisition stays open, the count grows, and the pile is eventually cut by whoever has an hour free.

Three numbers make the cap concrete for your team:

• Reader hours available this week, honestly, from the people who will actually read.
• Minutes per application at the quality you want. Four is realistic for a considered read. Ninety seconds is a skim, and a skim is a different decision.
• The advance rate your cut rule produces, which tells you whether one tranche can fill a first round.

Multiply the first two, compare against the third, and the window sets itself.

Pausing and closing are different mechanically, and the difference matters to the candidate. Unpublishing keeps the requisition and the pool intact while stopping new arrivals, which is what you want between reading tranches. Closing ends the round and owes everyone an answer. Whichever your system calls it, check what a candidate sees when they follow an old link, because a posting that returns a blank page or a generic careers listing is the version most people will hit.

What Does the Two-to-Four-Week Rule Assume?

That applications arrive slowly and evenly, so time is a proxy for reach. It was a reasonable proxy while a posting drew a handful of applications a day and the risk was closing before enough people saw it. One-click and agent-assisted applying broke the proxy in one direction only: reach got faster, and reading capacity did not move at all.

The benchmarks confuse two different things. Time-to-fill measures the whole hire, not the application window. SHRM's 2021 benchmarking data, collected from its own member organizations, puts median time-to-fill for nonexecutive roles at 44 days, with the median organization spending 5 days screening applicants and 7 days interviewing ¹. That median sits inside an interquartile range of 28 to 73 days, and the mean is 54, so it describes a spread.

Vacancy-duration measures are a different quantity again, built on JOLTS and counted in working days where SHRM counts calendar days, which is why numbers from the two families should never be compared directly ². Neither one tells you how long to accept applications. They tell you how long the process takes once someone is reading, and the reading is the part you control.

The calendar keeps one honest use. A licensed, niche or geographically constrained role can genuinely take weeks to reach the few dozen people who qualify, and a volume cap will never bind there. Check the arrival curve in the first 48 hours to see which of those you are in.

Say the Limit in the Posting

One line, in plain words, near the top: this posting closes when it reaches 150 applications or on the 12th, whichever comes first, and everyone who applies gets an answer either way. A candidate who reads that and applies has agreed to the terms. A candidate who arrives after the close sees a closed posting and a reason, which is a materially different experience from the one most people are having now.

The evidence on rejections points the same direction: what people react to is the absence of an explanation more than the identity of the decider. In a vignette experiment with 921 working-age Austrians, an automated rejection with no explanation rated lowest on every outcome measured, while an explained one rated the same as a human rejection without an explanation ³. Those are hypothetical scenarios and fractions of a point on a five-point scale, so do not oversell it. The direction is still the useful part: state the reason, even when the reason is capacity.

The wording matters less than the keeping. If you publish a cap and then leave the posting up for another month, you have taught candidates that your postings lie, which is worse than never publishing one. A req that hit its cap and closed in two days is the failure mode to plan for before you publish the number.

When Do You Reopen a Requisition That Did Not Produce a Hire?

When you have changed something. Reopening an unchanged posting draws a similar pool, often literally the same people, and burns another week of reading. Change the pay band, the location rule, the seniority, or the ask in the application, then reopen and say what changed. A rotation of two-week windows with a change between each is a better instrument than a permanently open requisition.

What to change, in rough order of effect: the range, if the last pool priced you out of the market. The level, if the applications were consistently junior against the rule. The first-project description, if the strongest candidates withdrew after the phone screen. The application ask, if completion was near zero.

Between windows, close the posting properly rather than hiding it. An open requisition nobody is reading collects applications from people who are counting on an answer, and it degrades the response-rate number your team is judged on. The posting changes that reduce generated volume is what to spend the gap on, and why applications per opening rose this far explains why the next window will fill faster than the last one.

Go back to the last pool before you spend on a new one. The candidates who nearly cleared the rule are already read, already interested, and cost nothing to contact, and a short note saying what changed since they applied lands better than a fresh posting ever does. Do that first, then reopen.

See a sample report >