Settle Four Things at Intake, Then Test Them on Three Real Profiles

What has to be agreed before the req opens?

Two or three competencies that separate this hire from an adequate one, the stage that will produce evidence for each, what the manager will accept as sufficient, and who decides when the loop splits. Those are the four, and the circulating intake templates cover none of the last three. They cover backfill status, org placement, must-haves and the band, which is a specification for reading resumes.

The reason the last three matter is that nothing downstream will correct them. In LinkedIn's 2025 survey of 1,271 recruiting professionals at management seniority or higher across 23 countries, only 25% reported feeling highly confident in their organization's ability to measure quality of hire effectively, while 89% agreed it will become increasingly important ¹. That is a self-report from a convenience sample of talent professionals, not an audit of what any employer measures, and LinkedIn sells hiring products, so read it as practitioners describing their own confidence. Read that way it still settles what this meeting has to do. When the people who run hiring are that unsure of the six-month number, the agreement written today has to carry its own check.

The fourth item looks bureaucratic until the week it is needed. Settle who decides when the loop splits while there is no candidate to argue about, because after the split it reads as one person overruling another. It is the same question as who gets to decide what qualified means, asked early enough to be answered calmly.

Name the stage that produces the deciding evidence

Ask which stage will produce the evidence the hire rests on, and write the answer in the intake doc. Most searches never answer it, so the decision settles wherever the loop happens to generate a strong impression, and the earlier stages get designed as filters feeding a stage nobody named. Naming it changes what gets built: the deciding stage earns the rubric and the time, and everything upstream can be cheaper.

The funnel already shows where the deciding is not happening. In one applicant-tracking vendor's dataset, weighted to venture-backed technology employers, recruiter screens pass roughly 35% of candidates who reach them, while later stages convert at 95% post-onsite and 81% at offer, and around 13% of hires included a take-home component ². Those are passthrough rates among candidates who reached each stage, so they do not multiply into anything end to end, the heaviest filtering happens earlier still at application review, and stage names are configured per employer inside the tool. The shape is the useful part: by the time a loop reaches its late stages it is confirming, not choosing.

So choose it on purpose. Most loops inherit their deciding stage from whatever the last search happened to do. Which stage actually carries the decision is the design question, and what happens to stage pass rates once candidates are working with AI is what makes last year's answer unreliable this year.

Calibrate on three real profiles, not on a description

Pull three profiles from the live pool with the names removed and have the manager react to them in the meeting. Twenty minutes of that beats an hour of adjectives, because the gap between the stated must-haves and the people a manager actually likes is the whole risk of the search, and it is invisible in a requirements list. Do it before sourcing starts, while the list is still cheap to change.

The reason this pays now is that review attention has become the scarce input. Across over 109 million applications and 247,000 jobs from January 2021 through March 2026, the same vendor reports the average recruiter processing 291 applications per hire against roughly 100 in early 2021, and the share of applications resulting in an interview falling from about 7-8% in 2021 to between 3.6% and 4.7% depending on role type ³. The numerator counts application records inflated by duplicate and automated submissions, and 291 is a ratio of two organization-level totals, so it is not any individual applicant's odds. What survives the caveats is that a vague requirements list no longer costs a few wasted screens. It costs the review capacity the search actually runs on.

Fix that capacity in the same meeting. How fast the manager will return feedback, how many interviewer hours a week the loop can absorb, and who covers a week of leave are all binding constraints, and none of them appear on an intake template. If the role's day-to-day work has changed since the last time this job was filled, finding out what AI actually does in the role belongs here too, before the requirements get written from memory.

When should you reopen the agreement?

Once, deliberately, after the first handful of candidates, and after that only if the role itself changes. Treat a changed mind as information rather than as a failure to be litigated: a manager who has now met six real people knows something nobody could have known in week one. What is not acceptable is the unannounced version, where the bar moves and only one person knows it moved.

Schedule that check against your own pipeline. Published benchmarks are the wrong clock for it. SHRM's 2026 benchmarking brief, built on data from over 4,600 organizations, reports a median time-to-fill of 39 calendar days for nonexecutive positions ⁴. That median runs across organizations that answered a survey, so an employer making two hires a year weighs as much as one making two thousand, and the respondents are self-selected members of a US HR association. Set it beside the same applicant-tracking vendor's median time to first fill, 56 days for business roles and 76 for technical ones ³, and the two figures differ for two reasons at once: the clocks differ, and so do the populations. Nothing published apportions the gap between those causes, so comparing your search to either one without matching both is the most common error available with the number.

A better trigger is internal and it is already sitting in the pipeline: once four or five candidates have been through the deciding stage, put the write-ups in front of the manager and ask whether the agreement still describes what they want. Doing that also settles the loop's shape while it can still be changed, which is the neighboring decision in who belongs on the interview loop. Whatever comes out of it goes back onto the same page, dated, so whoever inherits this search can see what changed and when.

See a sample report >