The Unmeasured Hire: Executive Cost-per-Hire Jumped 42% in a Year. Four in Five Companies Never Check the Return.

The median US company now spends $15,000 to fill an executive seat. A year ago it spent $10,600. The seat does not get filled any faster — executive time-to-fill has not moved a single day. And 80% of those companies will never measure whether the person they bought was worth it.

The numbers come from SHRM's 2026 Recruiting Executives Benchmarking brief — 4,657 respondents, fielded from November 24, 2025 to January 23, 2026, with directors and above deliberately oversampled. It is the closest thing the profession has to a public operating ledger, and read in sequence it describes a market that got materially more expensive without getting one degree more accountable.

What follows is that spend traced through a calendar, because the shape of the problem only shows up in sequence.

Day 0 — the requisition opens

Somebody signs off on a leadership seat. In the median organization, this seat will be filled from outside: SHRM puts the median share of executive positions filled externally at 100%. That is not a rounding artifact of small companies. Among organizations with 500–4,999 employees, external fill of executive roles moved from a median of 75% last year to 100% this year. Among organizations above 5,000 employees, it moved from 50% to 100%.

In one year, large and extra-large employers stopped promoting into executive seats at the median. SHRM says plainly that this shift "could be driving the rising cost-per-hire for executive positions," and it is hard to argue: buying is more expensive than building, and everyone decided to buy at once.

Day 45 — the offer is accepted

Median executive time-to-fill in 2026: 45 calendar days. In 2025: 45 days. Unchanged.

Now put the nonexecutive line beside it. Nonexecutive time-to-fill fell from 44 days to 39, and SHRM attributes the compression to AI: recruiting is the function where HR most commonly deploys AI tooling, and organizations with the most effective recruiting practices fill roles five days faster than everyone else.

So the volume end of the funnel sped up and the executive end did not. That is a diagnostic, not a coincidence. AI compresses screening throughput — reading, ranking, scheduling, coordinating. If executive searches had been bottlenecked on throughput, they would have compressed too. They did not, because the executive bottleneck is judgment: whether this person fits this company, this manager, and this phase of the business. No amount of screening speed touches that. We have written before about the hiring math nobody runs, and this is its cleanest confirmation to date — the pile got faster to sort, and the answer did not get easier to find.

The part of hiring that automation could fix got 11% faster. The part it cannot fix did not move at all.

Day 46 — the ledger closes

Cost-per-hire is booked. It includes, by SHRM's own definition, third-party agency fees, advertising, job boards, referral costs, travel, relocation, recruiter pay and benefits, and applicant-tracking systems, divided by hires.

Read that list again for what it excludes. It excludes the leadership hours consumed by the search. It excludes the cost of the seat sitting empty, which we have priced separately. And it excludes, entirely, the possibility that the hire does not work — a cost that dwarfs the fee and which we have put numbers to elsewhere.

Cost-per-hire is an input metric. It measures what you spent, and it closes on the day the offer is signed. It has nothing to say about what you got.

The premium multiple has nearly quadrupled

Here is a number SHRM does not print, which falls out of its own table. Divide the executive median by the nonexecutive median in each year:

2017: $5,000 ÷ $1,600 = 3.1×
2026: $15,000 ÷ $1,300 = 11.5×

Filling an executive seat cost roughly three times a regular hire nine years ago. It costs roughly eleven and a half times a regular hire now. Treat this as directional rather than precise — SHRM refines its benchmarking methodology between waves and the respondent base shifts year to year — but the direction is not ambiguous, and it is not explained by inflation. Nonexecutive cost-per-hire is flat in nominal dollars across the same nine years. The entire divergence sits on the executive line.

Month 6, Month 12, Month 18 — nothing happens

This is the part of the calendar with no entries in it.

Only 20% of organizations measure quality of hire. One in five. Unchanged from 2025. Among those that do, the instruments are unglamorous and entirely available to everyone else: performance appraisal scores, retention rates, and structured post-hire interviews.

And SHRM reports the payoff in the same brief: organizations that measure quality of hire report improved organizational fit among their new hires. Not improved technical output — fit. The act of closing the loop changes what gets selected for, because you cannot optimize toward a result you have never scored.

Set the two facts side by side and the year reads like this. Executive acquisition spend: up 42%. Executive time-to-fill: unchanged. Executive external fill: up to 100% at the median in large organizations. Share of organizations checking whether any of it worked: 20%, flat.

Why the loop stays open

The convenient explanation is that measurement is hard. The data says something less flattering.

Median requisitions per recruiter rose from 20 to 25 in a year. In extra-large organizations, the median went from 60 to 100 — a 67% increase, at organizations carrying a median of 0.8 HR professionals per 100 employees.

A recruiter carrying a hundred open requisitions cannot close a quality-of-hire loop. Not through incompetence — through arithmetic. Closing the loop means knowing, eighteen months later, what each of those hires was supposed to accomplish, who they reported to, what the organization actually needed at the time, and whether the person is doing it. Multiply by a hundred and the work does not exist. So the metric that survives is the one that can be computed on the day of signature and divided by a denominator. Cost-per-hire survives because it scales. Quality of hire does not scale, so it is not measured, so nothing corrects.

Note also the shape of the 20%: 23% of small organizations measure quality of hire, 23% of extra-large, and only 19% of midsize and 16% of large. Small companies match the largest, best-resourced enterprises on this — with none of their analytics infrastructure. Measurement is not, in fact, an infrastructure problem. It is a volume problem, and the smallest organizations do not have the volume to hide behind.

The metric that survives is not the important one. It is the one that survives contact with a hundred requisitions.

This is the whole argument for depth over volume, stated in someone else's data. A firm running a hundred searches optimizes for the number it can produce at that scale. A firm running ten optimizes for the number that matters, because it can still see all ten at month eighteen. That is not a philosophy. It is a constraint, and it is visible in SHRM's tables.

The Return Ledger

If you are going to spend $15,000 in hard cost and several times that in leadership hours, three entries written at offer — not at review time, when memory has reorganized itself around the outcome — convert an unmeasured hire into a measured one. Together they are the Return Ledger, and they take about forty minutes.

1. The month-18 sentence. One sentence, signed by the hiring executive before the candidate accepts, describing what will be observably true if this hire worked. Not a scorecard of objectives — those get renegotiated. One sentence about the state of the business or the function. Sealed, dated, and reopened at month eighteen.

2. The named alternative. Who was the internal candidate, and what specifically were they missing? If the honest answer is "there wasn't one," that is a finding about your pipeline, not about the market — and it belongs on the ledger, because at 100% external fill it is the single most expensive line item you are not tracking. It is the same gap we traced in the missing middle rung.

3. The fit thesis, in writing. Who this person reports into and how that person actually behaves in disagreement. Whether the company needs this seat to transform something or to multiply something that already works. What the seat requires at month 24, not month one. Written down, it can be scored later. Unwritten, the eventual post-mortem will conclude the candidate was "not the right fit" — a sentence that explains nothing and prevents nothing, and which is only ever produced by organizations that never specified fit in the first place. This is the 89% that the résumé cannot see.

Three entries. Reopened once. That is the entire difference between a $15,000 expense and a $15,000 investment with a known return.

What boards are doing instead

There is a revealing pattern one level up. Russell Reynolds' Global CEO Turnover Index found that in Q1 2026, 26% of incoming CEOs globally had already been a public-company CEO — up from 17% in Q1 2025 and 8% in Q1 2024. In the S&P 500 the figure hit 41%, the highest first-quarter level in nine years of tracking.

Boards are buying proof of prior performance at a rate that has more than tripled in two years. That is a rational response to uncertainty, and it is also a confession: when you cannot assess fit, you buy the closest available proxy for it, and pay for the privilege. The proxy is expensive and it is thin — 86% of 2025 CEO appointments across those indices were still first-time CEOs, so the strategy does not scale even at the top of the market. Below the S&P 500, in the founder-led and PE-backed mid-market where ETHOSLINK works, it does not exist as an option at all.

Which leaves the unfashionable alternative: specify the fit, hire against the specification, and score it at month eighteen. The spend has already gone up 42%. The measurement is free.

And it is worth remembering that the ledger opens on day one, not at month eighteen — most of what goes wrong is already in motion during the first 90 days after signing.