Per-Candidate Pricing Buys Volume When You Need Depth

Measure cost per hire, not cost per test

Divide by hires, not by candidates. An assessment at forty dollars a head sounds cheap until it runs on every applicant: three hundred applicants for one opening turns forty dollars into twelve thousand, spent before anyone has read a word of it. The same budget covers real reading time on the twelve people who reached a final round, and produces a record of what each of them did.

The denominator matters because the reference number is smaller than most teams assume. SHRM's benchmarking data puts median cost-per-hire at $1,244 for nonexecutive roles against a mean of $4,683, with a 25th percentile of $354 and a 75th of $4,375 ¹. The distribution is heavily right-skewed, so any single number misleads, and these are 2021 figures from SHRM member organisations, a sample that skews toward employers large enough to have an HR function.

Read what that definition includes before using it as a budget line. SHRM sums agency fees, advertising, job fairs, job boards, referrals, travel, relocation, recruiter pay and talent acquisition system costs ¹. It does not include the hiring manager's or the interviewers' hours, which is usually the largest real cost of a loop. So a tool whose pitch is that it saves interviewer time is being measured against a denominator that never counted that time.

Run the arithmetic on the last three requisitions before the demo. Applicants received, candidates assessed, hires made. Above roughly thirty applicants per hire, a per-candidate price gets expensive fast. In single digits, a subscription is the thing being overpaid for.

What does each pricing model assume about your volume?

Every model is a bet on how many people will take it. A per-candidate fee assumes hundreds, and at hundreds the marginal cost of one report has to sit near zero, which usually means software produces it. A per-seat subscription assumes a recruiting team running the tool constantly. A credit pack assumes bursts. A per-report price with a human inside it assumes a shortlist. None of these is dishonest. Each one names the volume the product was designed around.

The awkward part is that the method with the best evidence behind it is the one that costs human hours. In the corrected 2022 estimates, structured interviews rank first at .42 as a correlation with supervisor-rated performance, ahead of job knowledge tests at .40, work samples at .33 and cognitive ability tests at .31 ². The gaps at the top are small and the credibility intervals overlap, so read the order as a rough grouping. What is not rough is that a structured interview cannot be bought by the unit: it is question writing, a rating scale and interviewer time, and the corrected table is worth reading before a budget meeting.

That is the trade hiding inside the pricing page. Priced to run on everyone, the output is produced by software, so the artifact is what gets judged. Priced for a shortlist, someone reads it, so the reasoning can be judged. Which is worth paying for depends entirely on how many people reach the stage where reasoning decides anything.

Ask which stage the money is buying. Most teams need both, something coarse at the top and something legible at the bottom. Building the second one in-house is a real option, and it moves the cost into answer keys and reviewer hours.

Where does a credit plan stop being cheap?

At the overage line, where a credit plan stops being a plan and becomes a per-test price. Read three numbers on the order form: credits included, price per credit beyond, and when unused credits expire. The crossover is arithmetic you can do in the meeting: divide the difference between the two annual prices by the overage rate, and that is how many extra candidates it takes before the cheaper plan is the dearer one.

Two details move it more than the headline rate. The first is when a credit is consumed. A credit charged when the invitation is sent bills every candidate who never starts. A credit charged on completion bills only finished work, which is a materially different contract at the same nominal price.

The second is expiry. Annual credits that reset unused convert a variable cost into a fixed one and punish exactly the pattern most teams have: three months of heavy hiring and nine months of almost none. Monthly credits that reset are the same problem at higher frequency.

There is a timing argument too. SHRM's benchmarking data puts median time-to-fill at 44 days for nonexecutive roles, with a 25th percentile of 28 days and a 75th of 73 days ¹. That is calendar days from requisition to accepted offer, and it means a single hire routinely spans two credit months. Ask the vendor to model last year's actual volume, and to put the overage price in the order form where it binds.

Ask what the price buys in human reading time

Ask it as a number: how many minutes of a qualified person's attention does one unit of this price include? If the answer is zero, what you are buying is software, and it should be budgeted and argued about like software. If the answer is thirty or sixty minutes, the price is buying a reader, and the next questions are who that reader is and whether what they produce is a sentence or a number.

There is a floor under this that no pricing model escapes, and free sits below it. In a controlled test of unstructured interviewing, predictions of a student's grades made after conducting an unstructured interview correlated .31 with the outcome, while predictions from the prior record alone correlated .65 ³. Adding the conversation made the forecast worse. That was undergraduates predicting classmates in a lab, so the size of the effect does not carry over to hiring. The mechanism does: low-diagnostic information dilutes good information. A free unscored chat is a price too, and it sometimes buys negative information.

Three questions get most of the way through a pricing call. What is the unit, and when is it counted? What does a human do inside that unit? And what happens to the price at twice this volume and at half of it?

Then price the plumbing separately, because it is often where the real cost sits. Whether the tool needs to talk to your ATS at all usually has a cheaper answer than the one on offer, and it is worth settling before the per-candidate rate gets negotiated.

See what gets scored >