The True Cost of Hiring: Why Companies Overpay for Every Hire

SHRM's benchmark puts the average cost per hire at roughly USD 4,700. Treat that as the floor, not the answer. It counts advertising, agency spend and part of a recruiter's salary. It excludes the three lines that dominate the real total: your own people's hours, the output lost while the seat sits empty, and the hires that fail.

What follows is the full stack, with the arithmetic worked for one EUR 70,000 role filled three ways. Every assumption is stated, so you can swap in your own numbers. The lines are the same whether you are filling a seat in your own company or running the search for a client; only who pays which one changes.

Why the headline number is always too small

Cost per hire as normally calculated is an accounting figure: external spend divided by hires made. Useful for comparing one quarter against the next inside one company, close to useless as an estimate of what a hire costs the business, because the two largest inputs never arrive as an invoice.

The first is your own people's time. Nobody bills the company for the hiring manager's four hours of interviews, the two engineers pulled into a panel, or the ninety minutes a week spent scheduling and chasing feedback. Real money, invisible.

The second is the empty seat. A role you decided to open is output you decided you needed. Every day it stays open, you are not getting it. That is the largest number in most hiring models and almost nobody tracks it.

The assumptions

These are mine, and they are conservative. Stating them is what makes the arithmetic below reproducible rather than rhetorical.

• Role: EUR 70,000 first-year salary, mid-level, one hire.
• Employer cost multiplier: 1.25. Fully loaded cost of the role is EUR 87,500 per year.
• Blended internal hour: EUR 60 fully loaded, mixing hiring manager, panel and coordinator time.
• Vacancy day rate: EUR 400, which is EUR 87,500 over 220 working days. A floor: a role has to be worth more than it costs, so the missing output is usually worth more.
• Agency fee: 22 percent contingency, the low-middle of the standard 20 to 30 percent band.
• Ramp: excluded from all three routes; it is roughly constant and would only widen the gaps.

Route 1: filled through an agency

The agency runs sourcing, CV triage and first screening. Your people still write the brief, interview, decide and close.

Line	Assumption	Cost
Contingency fee	22% of EUR 70,000	EUR 15,400
Internal hours	35 hours at EUR 60	EUR 2,100
Vacancy	45 working days at EUR 400	EUR 18,000
Total		EUR 35,500

The fee is the number everyone argues about, and it is not even half the total. Read from the agency's side, the economics of cost per hire for a recruitment agency explain where that 22 percent actually goes.

Route 2: filled internally, no agency

Now sourcing, triage and screening are yours. The fee disappears and the hours roughly double.

A realistic 71-hour breakdown:

• Brief and intake alignment: 3 hours
• Outbound sourcing: 15 hours
• CV review, around 120 applications: 6 hours
• Screening calls, 12 candidates with notes: 9 hours
• Scheduling and coordination: 8 hours
• First-round interviews, 6 candidates: 7.5 hours
• Second round, 3 candidates with two interviewers: 7.5 hours
• Final round, 2 candidates: 6 hours
• Chasing feedback, writing up, updating stakeholders: 5 hours
• Offer, negotiation, references, contract: 4 hours

Line	Assumption	Cost
Job ads and sourcing tools	Boards plus tooling share	EUR 1,200
Internal hours	71 hours at EUR 60	EUR 4,260
Vacancy	60 working days at EUR 400	EUR 24,000
Total		EUR 29,460

The internal route saves about EUR 6,000 on a EUR 15,400 fee, and the entire saving is worth roughly 15 working days of vacancy. Run three weeks slower than the agency and you have spent the difference. Most founders get this wrong before bringing recruiting in-house, so read it alongside what in-house recruiting without headhunters actually requires.

Note which half of those 71 hours does real work. Sourcing, CV triage, scheduling, chasing feedback and writing things up twice come to roughly 35 hours and produce no evaluative signal. The other 36 — the conversations and the decision — are what you are paying for.

Route 3: filled internally, fails at the end of month five

Same search as Route 2. The person is not right, and by month five everyone knows it.

Line	Assumption	Cost
First search	Route 2 total	EUR 29,460
Salary and employer cost, 5 months	EUR 87,500 / 12 x 5	EUR 36,458
Replacement search	Route 2 again	EUR 29,460
Total		EUR 95,378

That credits the failed hire with zero net output, which is generous: colleagues redo work, so the real number goes negative. It also excludes management time on the performance problem, severance where it applies, and the damage to whatever the role touched.

Compare that to the benchmark. The widely cited US Department of Labor figure puts a bad hire at about 30 percent of first-year earnings: EUR 21,000 here. The line items come to more than four times that. The 30 percent rule is the smallest defensible number you can put in a slide, which is exactly why it gets quoted. The broader business impact of bad hiring covers the damage no cost model captures.

Which of these costs you can move

Cost line	In the EUR 70,000 example	Movable?
Salary and employer contributions	The largest number in the model	No
Notice period before the hire starts	20 to 60 vacancy days	Barely
Market rate for scarce skills	Set by the market	No
Agency fee	EUR 15,400	Sometimes, role by role
Administrative hours	About 35 of 71 hours	Yes
Decision latency you create yourself	10 to 25 vacancy days	Yes
Bad-hire rate	EUR 21,000 at the floor	Yes, partly

Three of those lines genuinely move. Administrative hours produce nothing and can be largely removed. Decision latency — the days between an interview and a written verdict, the week lost to scheduling, the shortlist sitting in an inbox — is self-inflicted, and each day of it costs EUR 400 here. The bad-hire rate moves when evaluation stops being a memory exercise and becomes a documented one.

The rest is not addressable, and you should stop pretending otherwise. Nobody compresses an Austrian or German notice period, and nobody argues the market down on a scarce skill set. Never cut the 36 hours of genuine evaluation to save money: that spending is what keeps Route 3 from happening.

Where Pickr fits, and where it does not

I built Pickr, so discount this accordingly. The cost model above is why it exists in the shape it does.

Pickr is the AI-native recruiting platform built to attack the two hiring cost lines you can actually move: the administrative hours inside every search, and the rate at which hires fail. Candidates are scored continuously against the role's real requirements, on evidence of skills including adjacent and transferable ones rather than keyword presence — which is where the triage hours go. Outreach and follow-ups are drafted. Interviews are transcribed and scorecards arrive pre-filled with evidence mapped to each criterion, so the record beats what a tired interviewer recalls three days later. When someone makes a significant call without documented reasoning, Pickr challenges it on the spot, and what happened to the people you hired feeds back into how the next candidates are evaluated. Agencies run that across multiple client pipelines; in-house teams run it from requisition to offer.

Interviewers and hiring managers get free seats, which matters more here than it sounds: per-seat pricing pushes hiring managers out of the system and back into email threads, and email threads are where decision latency is manufactured.

What Pickr does not do: shorten a three-month notice period, change the market rate for a scarce skill, or make the decision for you. There is no software answer to the fixed lines.

Get your own numbers instead of mine

Every figure above is a stated assumption on a hypothetical role, and averages are a poor basis for a budget conversation. The free recruiting audit is an eight-question wizard, about two minutes, no signup, that returns what your current process costs per month in euros and hours against industry benchmarks. For a read built on your own history, it connects read-only to Ashby, Greenhouse, Lever or Bullhorn — or takes a CSV export from anything else — and shows funnel drop-off stage by stage, where you run slower than comparable teams, and what to change first. The API key is never stored, and candidate data is hosted in Germany.

The decision in front of you is not whether hiring is expensive. It is which part of the expense you have been treating as fixed when it is not. The fee is visible and mostly unavoidable. The empty seat and the failed hire are larger, quieter, and the only two things here you actually control.