Most Hiring Cost Statistics Are Wrong. Here Is the Real Math on a Senior Engineer.

If you have ever built a hiring budget, you have probably quoted one of these:

• "A hire costs about $4,700."
• "Engineering roles take 62 days to fill, according to SHRM."

Both come from real SHRM reports. Both are being repeated wrong. I went and read the source PDFs, and the second one is not in them at all.

This matters more than pedantry. If you are deciding whether to hire in-house, use an agency, or contract someone, you are comparing options against a baseline — and that baseline is off by more than an order of magnitude in places.

The $4,700 figure is a mean hiding a very skewed distribution

SHRM's 2022 Talent Access Report publishes the whole distribution, not just the headline:

The famous number is that $4,683 mean. The median organisation spends $1,244 — the mean sits above the 75th percentile, which is what happens when a handful of enormous executive searches drag an average upward.

SHRM knows this. Their 2025 Recruiting Benchmarking Report reports medians of $1,200 and $10,625 and says plainly that it uses medians "to avoid the influence of extreme outliers." Their press release the same month led with means: $5,475 and $35,879. Two SHRM publications, one month apart, differing by about 4.5x, and the higher pair is the one that spread.

There is a second problem. SHRM defines cost per hire as recruiting-function spend divided by all hires. It excludes interviewer time and onboarding entirely, and it averages your expensive senior searches together with every cheap requisition you filled that year. It was never meant to price a single senior engineer, and it cannot.

"SHRM says engineering takes 62 days" is not a real citation

Here is the 2022 time-to-fill table: executives, median 60 days and mean 62. Nonexecutives, median 44 and mean 54.

The widely-quoted "62 days for engineering" is the executive mean, relabelled. The word "Engineering" does not appear anywhere in either report. SHRM publishes no breakdown by role or function at all, so no SHRM figure for engineering roles exists to cite.

While we are clearing out dead sources: the DHI-DFH index, still quoted in hiring posts today, was discontinued in April 2018. Anyone citing it is citing data that stopped being collected eight years ago.

What the good data says

The most credible time-to-fill numbers come from Ashby, because they are drawn from actual applicant-tracking records rather than surveys asking recruiters to remember — 54 million applications across 93,000 jobs. Their recruiting operations benchmarks break down the way you would want:

• Junior roles: 52 days median
• Mid-level: 63 days
• Senior: 71 days
• Technical roles overall: 75 days, against 60 for business roles

So a senior technical hire is a roughly two-and-a-half month process from opening the req, and that is the median — half of them take longer.

Their recruiter productivity data adds the part that never makes it into a budget: 24.7 interviewer-hours and 17.9 interviews per engineering hire. That is your existing senior engineers, the ones you are short of, spending three full working days each cycle interviewing.

Even "what does a senior engineer cost" has two right answers

Before you can price the hire you have to price the person, and the two most-quoted sources disagree by nearly $100,000.

The BLS Occupational Employment and Wage Statistics for Software Developers (SOC 15-1252, May 2025 — the most recent published) gives a national distribution:

Levels.fyi's 2025 report puts the median Senior Engineer at $312,000.

That gap is not one of them being wrong. It is three things stacked:

1. BLS excludes equity. Its wage definition covers base pay and commissions but not stock, RSUs, or nonproduction bonuses. For a senior engineer at a public tech company, that is often the larger half of the package.
2. The populations differ. BLS is an employer-reported probability sample covering every US employer of software developers. Levels.fyi is self-reported and skews heavily to big tech and the Bay Area, Seattle and New York.
3. Self-selection. People submit to Levels.fyi when they are benchmarking a strong offer.

The practical read: if you are a normal company hiring in a normal metro, the BLS 75th percentile is your realistic anchor. If you are competing with Meta for the same candidate, Levels.fyi is. Quoting the $312,000 figure at a Midwest manufacturer, or the $135,980 one in Palo Alto, will both lose you the hire.

Adding it up honestly

So take that $171,980 anchor — a senior developer at the 75th percentile of the national distribution.

The recruiter fee dominates everything else, and it is precisely the line that SHRM's $1,244 median hides. Staffing Industry Analysts' survey work puts the median contingency fee at 20%, with direct-hire specialists charging 20–25%.

Notice what is still missing: the cost of the role sitting empty for two and a half months. That number is specific to your business, and it is usually the biggest one on the page.

One statistic I am deliberately not quoting

You will see "a bad hire costs 30% of first-year earnings, according to the US Department of Labor" everywhere. I could not find the document. A search restricted to dol.gov returns nothing, and every citation traces back to a staffing-agency blog asserting it without a number, title, or date. I am not going to repeat it.

The defensible version is Boushey and Glynn's meta-analysis for the Center for American Progress — a real synthesis of 30 case studies, finding a median turnover cost of 21% of annual salary. Two caveats worth stating: the underlying studies run from 1992 to 2007, and the alarming 213% top of its range is an executive outlier, not a senior engineer.

The market is tighter than the headlines suggest

It is easy to assume that years of tech layoffs made senior engineers easy to hire. The data says otherwise, and the distinction is precise.

Software development job postings sit at 74.4 on Indeed's index against a February 2020 baseline of 100, while postings overall sit at 101.8. So the contraction is real and specific to software. But 71% of the recovery in software postings came from senior roles, and unemployment in computer and mathematical occupations was 3.1% as of August 2026, against 4.3% across the economy.

Research by Brynjolfsson and colleagues using ADP payroll data finds a roughly 19% employment shortfall for 22-to-25-year-olds in AI-exposed occupations, with no comparable gap for experienced workers.

Put together: the junior market is genuinely loose, and the senior market is not. If you are hiring the senior engineer, the layoff headlines do not describe your situation.

Which means vetting is the whole thing you are buying

Look back at that table and ask what the $34,000 to $43,000 actually purchases. Not sourcing — you could post the role yourself in an afternoon. Not the interview — your own engineers do that, for 24.7 hours. What the fee buys is filtering: someone else absorbing the cost of finding out that most applicants are not the person you need.

That reframes the build-versus-hire-versus-contract question. The bottleneck is not access to candidates. It is the cost of separating signal from noise, paid in your senior engineers' attention, over two and a half months, while the work does not get done.

There are three broad ways to pay that cost.

The middle row is where most bad outcomes live. An open marketplace looks cheap because the platform fee is small, but it has simply moved the filtering cost back onto you — and you are now doing it without a recruiter, on candidates you cannot easily verify.

Where a pre-vetted network fits

Toptal sits in that third row. The model inverts the usual order: screening happens before you ever see a candidate, and you engage on an hourly basis rather than paying a percentage of a salary.

I should be straight about which parts of that are verifiable and which are marketing. Toptal's well-known claim is that it admits roughly the top 3% of applicants. That is a self-reported figure. I have not seen it independently audited, and you should read it as positioning rather than a measured statistic.

What I can speak to directly is that the screening is real, because I went through it myself and I am in the network. It is not a formality, and it is not a resume check.

Here is the honest economic case, which does not depend on believing the 3%:

• You are not paying a placement fee. The $34,000–$43,000 line disappears. You pay an hourly rate instead, and that rate is higher than a local market average precisely because the filtering is priced into it. Whether that trade is good depends entirely on how long the engagement runs — for a defined project it is usually favourable, for a permanent role it eventually is not.
• You are not spending 71 days. For work that is already blocked, elapsed time is often the single largest cost, and it is the one nobody puts in the budget.
• You are not spending 24.7 hours of senior engineering time per hire on interviews.

Where it is the wrong tool

I would rather you not click through and then find out the hard way:

• For a permanent, core role, hire. A contractor does not accumulate institutional knowledge, and over a multi-year horizon the hourly premium stops making sense. Pre-vetted networks are for defined work with a shape you can describe.
• Technical vetting is not fit vetting. No screening process, however hard, tells you whether someone will work well in your codebase, your domain, or your team. That evaluation is still yours to do, and you should still do it.
• Classification is your problem, not the platform's. If you engage a contractor and then manage them exactly like an employee, you have taken on a misclassification risk regardless of who introduced you.
• If you can evaluate the work cheaply, you may not need the filter. For a small, low-stakes, easily-verified task, paying for someone else's screening is paying for something you could do yourself.

The short version

1. Stop quoting the $4,700. It is a mean over a skewed distribution, it excludes interviewer time and onboarding, and the median organisation spends $1,244.
2. Nobody has published a SHRM figure for engineering roles. If a vendor deck cites one, they made it up.
3. Use the real numbers: 71–75 days to fill, 24.7 interviewer-hours, and a placement fee of 20–25% of first-year salary that dwarfs every other line.
4. Decide which cost you are actually optimising — money, elapsed time, or your senior engineers' attention. They pull in different directions, and most hiring plans never say which one is binding.

All benchmarks here are 2025-vintage, the most recent published as of writing. If you are reading this much later, check whether the underlying reports have been updated.