The hidden cost of a bad hire: what we learned from 247 CVs
A bad hire doesn't cost the salary. It costs the time, the team's energy, and the missed opportunity.
A well-briefed HR director will tell you a bad hire costs between 6 and 12 months of fully-loaded salary. That's what HR textbooks teach. That's what the HRIS reports.
It's wrong. Or rather: it's well below reality.
Here's what we learned analyzing 247 CVs for a Senior Product Designer role at one of our first customers. And measuring what really happened during the 9 weeks of the process.
What you see when a hire fails
The candidate signs. Three months later, they leave — or they're let go. You calculate the visible cost:
- 3 months of fully-loaded salary: ≈ €18,000
- Agency fees (15% of package): ≈ €9,000
- Onboarding (training, equipment, mentor): ≈ €4,000
Visible total: ~€31,000.
That's what shows up in the P&L. That's what gets booked as "the cost of a bad hire". And that's what gets minimized with "it happens".
What never shows up in the P&L
Here's what we quantified by tracing back this specific case:
Manager time absorbed: 47 hours. The future Designer's manager spent 47 hours on the process — scoping calls, portfolio reviews, interviews, debriefs, salary negotiation. At €80/h fully-loaded: €3,760. Those 47 hours weren't on his plan. They pushed back a product release by three weeks.
Cost of the delayed release. The team of 4 designers + 6 devs operated below capacity during this period because the roadmap was blocked. Estimated opportunity cost on the delayed go-to-market of one feature: around €80,000 (revenue not captured + commercial re-engagement).
Team disengagement. During the 3 months when the new designer wasn't holding the role, two senior designers absorbed his backlog. One of them resigned the following month. Direct cost of replacing that designer: ~€70,000 (agency + 6 months of reduced productivity).
Real total: ~€185,000.
That's 6 times the visible cost.
Why a hire fails
We looked at the 247 CVs from the role. Here's the breakdown:
- 247 CVs received
- 89 CVs read in full by a human (35%)
- 41 CVs shortlisted on first read (17%)
- 18 candidates called (7%)
- 8 candidates interviewed (3%)
- 1 candidate signed (0.4%)
- 0 candidates kept after probation
Which link broke?
Not sourcing — there was volume. Not the interview — the 8 candidates seen were properly evaluated. The weak link is between CV 90 and CV 247. Those 158 CVs were never read. Nobody had the time. And statistically, in any distribution, 3 to 5 perfect candidates were in that unread pile.
The hire didn't fail at the interview. It failed at the reading.
What Maya did differently on the next role
For the next role — Senior Backend Engineer — the same team handed sourcing to Maya. Here's what we measured, side by side:
| Step | Designer role (human) | Backend role (Maya) |
|---|---|---|
| CVs scanned | 89 / 247 (35%) | 423 / 423 (100%) |
| CVs scored against brief grid | No | 423 |
| LinkedIn profiles analyzed in addition | 0 | 142 |
| Video pre-qualification (15 min) | 0 | 31 |
| Shortlist delivered | 8 candidates in 5 weeks | 6 candidates in 9 days |
| Manager hours absorbed | 47 h | 9 h |
| Hire still in role at 6 months | No | Yes |
Time-to-hire divided by 4. Manager hours divided by 5. And a higher 6-month retention rate, because the shortlist was wider at the source and matching was based on the entire pool — not on the 35% a human had time to read.
What Maya doesn't do
Maya doesn't decide who you hire. Maya doesn't run the final interview. Maya doesn't negotiate salary. Maya doesn't get the signature.
Maya removes the weak link: she makes sure the CV pile is read in full, that every candidate is scored against the brief, and that your manager sees the 6 best profiles — not the first 8 someone happened to open.
The rest is your job. Maya doesn't replace it. She gives it back to you.
The real ROI
People often talk about the ROI of a recruitment AI agent in terms of time savings. That's true, but it's secondary.
The real ROI is avoiding a bad hire.
For a team that hires 12 times a year, with an average failure rate of 20%, the hidden cost documented above represents around €440,000 in invisible annual losses. Halving that failure rate — which is the reasonable target of a Maya deployment — frees up €220,000 in cash that nobody had identified.
That's more than a Senior Designer's fully-loaded salary.
Ironic.