Cost Per Hire Counts the Recruiter and Skips the Panel

What belongs in cost per hire?

Cost per hire is external recruiting spend plus internal recruiting cost, divided by the hires made in the same period. The standard version of that definition is more specific than most teams assume, and its exclusions are where the argument lives. What it sums is recruiting spend. What it omits is the time of everyone in the process who is not a recruiter.

SHRM's benchmarking definition sums third-party agency fees, advertising, job fairs, job boards, referral costs, applicant and staff travel, relocation, recruiter pay and benefits, and talent acquisition system costs, divided by the number of hires ¹. Read that list again and notice who is missing: the hiring manager, the panel, whoever wrote the exercise and whoever graded it. Under this definition an interview loop can double in length and the metric will not move at all.

The figures attached to that definition carry a warning most quotations drop. In the same dataset, the median for nonexecutive roles was $1,244 across 472 organizations, with a 25th percentile of $354, a 75th percentile of $4,375 and a mean of $4,683, while executive roles ran at a median of $8,750 against a mean of $28,329 ¹. A median that far from its own mean signals a long tail, and any single number pulled from that distribution misleads in whichever direction it gets quoted.

Check the vintage of any average before quoting it. The figure most often repeated as the average cost per hire comes from a SHRM report published in 2016 covering fiscal 2015, a different survey from the medians above, and it still circulates as though it described this year ¹.

Why can't you benchmark this number outward?

Because the boundary is set per company and nobody publishes theirs. Which requisitions counted, whose salaries counted as recruiting cost, whether an internal transfer is a hire, whether a contract conversion is a hire: each of those is a local choice, made once and rarely written down. The arithmetic is standard. The boundary is not.

Role mix alone breaks the comparison. In that same benchmarking data the executive median sits at $8,750 against $1,244 for nonexecutive roles ¹. An employer whose year held three director searches and an employer whose year held thirty individual-contributor hires can run equally efficient processes and report figures a multiple apart, with neither of them wrong about anything.

Look at who was surveyed. Those medians come from a benchmarking survey of the member organizations of a US HR membership association, and they describe 2021 ¹. Membership of a professional body skews toward employers large enough to run a formal HR function, which is a narrower set than employers in general, and the data predates most of what changed in application volume since.

That leaves the outward comparison useless and the metric itself intact, which is a narrower and more actionable complaint. Your own figure, computed on a boundary you wrote down and held for four quarters, tracks something real about your process. The smallest hiring process a very small company can defend is where the boundary is easiest to write, because nearly everyone in the process is visibly doing another job as well.

Track three internal lines, not one

Split internal cost into recruiter hours, interviewer hours and tooling amortized per hire, then report the three beside the total. They move in different directions now, so a single internal number can hold steady while its composition inverts underneath it. A budget defended with one line will not survive the first follow-up question.

The recruiter line is under load, and the direction is documented even where the level is not. SHRM's 2026 recruiting benchmarking brief reports that extra-large organizations experienced a 67% increase in median requisitions per recruiter, and that over 2 in 3 organizations reported struggles hiring for open positions that year ². That is a change in a median rather than a workload level, it covers one size band, and the brief attributes no cause to it, so do not supply one.

The volume behind those hours is documented separately. One applicant tracking vendor's customer base, which skews to venture-backed technology employers, now averages 291 applications per hire against roughly 100 in early 2021 ³. Review time scales with the number of records opened, which is the mechanism by which a cost line grows while the hire count stays flat.

• Recruiter hours. Price them against the applications actually reviewed. The gap between that count and applications received is the part of the workload nobody is paying for yet.
• Interviewer hours. Outside the standard definition entirely ¹, and usually the most expensive hour of the interview loop. Count them separately or lose the argument about loop length permanently.
• Tooling per hire. SHRM's 2025 talent trends round names recruiting as the HR area using AI most, concentrated in writing job descriptions (66%) and screening resumes (44%) among organizations already using AI in HR ⁴. That spend lands on the recruiter line, so it can lower one internal line while leaving the interviewer line untouched.

Whether the review load is better answered by an exercise than by more reading is a separate question with its own arithmetic: is it cheaper to assess every applicant than to screen resumes at all.

Report last year again on this year's boundary

Before presenting either number, recompute last year's cost per hire using this year's boundary and put the two restated figures side by side. That is the only honest comparison available to you, and it takes an afternoon. A year-over-year change computed across two different boundaries is an accounting artifact wearing a trend line.

Write the boundary as four sentences at the top of the report: which requisitions are in scope, whose hours count as recruiting cost, what counts as a hire, and which period a hire lands in. Then leave it alone for four quarters. The urge to improve it arrives every time somebody spots an omission, and every improvement resets the series to nothing.

One external comparison is worth making, and it is not against a benchmark. Price the same role two ways: what it cost you, and what a contingency agency fee would have cost at that salary. Agency fees sit inside the standard definition ¹, so the counterfactual is computed on the same footing, and it answers the question a finance partner is actually asking, which is whether the internal function earns its own cost.

What this usually surfaces is uncomfortable. The pipeline with the lowest cost per hire is often the one consuming the most senior time, because senior time was never in the formula. Say that out loud when you present the number, and put the interviewer-hours line on the same slide so the room sees both halves at once. How many people you actually need next year is the conversation this figure feeds, and it deserves the version with the hours in it.

See a sample report >