You hired someone who looked perfect. Strong interview. Articulate. Resume coherent. Three months in, it is clear it will not work. Performance below expectations, friction with the team, missed deadlines. You need to let them go, reopen the search, and start over.
What did that just cost you? The honest answer surprises almost every founder and small-business owner who runs the math for the first time. It is also worth being honest about what is measured and what is estimated, because most of the numbers you will see quoted are softer than they look.
What the evidence actually says
The most rigorous public figure comes from the Center for American Progress. Heather Boushey and Sarah Jane Glynn (2012) reviewed the published academic literature on turnover costs and found 11 empirical papers providing 31 separate case studies. Across roles paying under 75,000 dollars a year, the median cost to replace an employee was about 21% of annual salary 1. That is the cleanest, most transparent number on this question, and it is built on peer-reviewed case studies rather than a vendor survey.
Two honest caveats. First, the 21% measures the cost of replacing someone who leaves: recruiting, selection, onboarding, and the productivity ramp of their successor. It does not capture the ongoing cost of a bad hire who stays, underperforms for months, and is eventually let go. That person costs more. Second, the figure is conservative by design and excludes higher-wage and executive roles, where the underlying studies run considerably higher.
For a number that specifically models a bad hire rather than general turnover, the UK’s Recruitment and Employment Confederation calculated that a poor mid-level hire on a 42,000 pound salary can cost more than 132,000 pounds, over three times annual salary, once wasted pay, training, lost productivity, and re-hiring are totaled 2. That is a modeled estimate, not a measurement, and it relies on a contested input, so treat it as a credible ceiling rather than a fact.
A figure you should ignore: the “30% of first-year salary, U.S. Department of Labor” line repeated across the HR industry. We tried to trace it to a primary DOL document and could not. Every secondary citation we found gave no title or link, and a direct search of dol.gov returns nothing matching it. We do not use folklore.
The peer-reviewed academic record is narrower than the marketing copy implies. Meta-analyses confirm the direction (Park and Shaw, 2013, found turnover negatively correlated with organizational performance; Hancock et al., 2013, found the same with a weaker coefficient) 3 4, but they report correlations, not dollar costs. No randomized study puts a clean causal price on a single bad hire. So the honest framing is this: replacement alone averages around a fifth of salary, the full cost of a bad hire who stays is credibly higher, and anyone quoting a precise universal number is estimating, not measuring.
A transparent model for an 80,000 dollar role
No study measures your exact role, so here is a line-item model you can adapt. A mid-level individual contributor: 80,000 dollars base, at-will terms, departs at month six.
Direct separation costs:
- Severance (no federal mandate; many firms pay 1-2 weeks per year of service plus accrued PTO): about 3,000 to 4,500 dollars
- Final accrued PTO payout: about 1,500 to 3,000 dollars for half a year
- Benefit continuation administration overhead: about 200 to 500 dollars
Subtotal: about 5,000 to 8,000 dollars.
Cost of the original hire, already spent and now wasted:
- Recruiter or sourcing time (internal HR, founder, or external): typically 30 to 60 hours of senior time, valued at 3,000 to 7,000 dollars
- Job board postings, ATS overhead, background checks: 500 to 2,000 dollars
- Onboarding setup (laptop, accounts, training materials, ramp curriculum): 2,000 to 5,000 dollars
Subtotal: about 5,500 to 14,000 dollars.
Indirect cost during the six months on payroll:
- Salary paid during ramp at below-expected productivity. A common convention assumes a new mid-level hire reaches roughly 60% productivity for the first 3 months. That is roughly 8,000 dollars of salary paid for output that did not materialize.
- Time the rest of the team spent absorbing uncovered work. Conservatively 10-20% of two coworkers’ time over those months: about 8,000 dollars.
- Manager time on 1:1s, course corrections, performance documentation, and the difficult conversations before separation: about 5,000 to 10,000 dollars.
- Deteriorated morale and the hidden tax of “we hired wrong again”: real, not counted here.
Subtotal: about 21,000 to 26,000 dollars that is observable.
Cost of the next hire:
- The recruiting cycle again: 3,000 to 7,000 dollars
- Onboarding setup again: 2,000 to 5,000 dollars
- The same ramp problem with the next person: 60-90 days at ~60% productivity, roughly 8,000 dollars of salary for partial output
Subtotal: about 13,000 to 20,000 dollars.
Reconciling the model with the evidence
Summing the midpoints lands around 50,000 to 65,000 dollars for an 80,000 dollar role that lasted six months, roughly 62-81% of annual salary. That is higher than the Center for American Progress 21% median, and the reason is exactly the caveat above: the 21% measures replacing someone who leaves cleanly, while this model includes the months of underperformance, team drag, and manager time of a hire who stayed and failed first. The 21% is the floor; a bad hire who stays sits above it.
For revenue-tied roles (sales, customer success, account management) the math gets worse, because the cost includes revenue that was never generated, not just money spent. None of these are measured facts about your business. They are a transparent model, so you can replace the inputs with your own and see the shape of the number.
Why this keeps happening
The structural reason is almost universal across small and mid-size businesses: most hiring decisions are made on unstructured interviews, scored from impression rather than evidence. For businesses without a dedicated HR function, the minimum viable process for hiring without an HR department covers the discipline needed to avoid these failure modes. Each candidate answers different questions, gets evaluated against the interviewer’s implicit criteria, and the final decision forms from fragments of memory.
Decades of organizational-psychology research are clear on what that costs in predictive accuracy. The McDaniel et al. (1994) meta-analysis 5 found structured interviews substantially more predictive than unstructured ones (mean validity around .44 vs .33, reaching .63 in the strongest panel-consensus conditions). Schmidt and Hunter’s 1998 synthesis of 85 years of selection research 6 reported .51 vs .38. Wingate et al. (2025) re-validated the direction with modern data 7. Translated: an unstructured interview explains less than 15% of the variance in on-the-job performance. A memory-based hiring decision is closer to a coin flip than to a measurement.
Without structure, two failure modes repeat:
The Performer. The candidate knows how to sell themselves. The right things, in the right tone, at the right pace. The interviewer believes them. The person cannot or will not deliver once on the team.
The Missed Diamond. The candidate is genuine and has what the role needs, but communicated poorly, was nervous, or did not look the part. They get filtered out and go to a competitor. This error is invisible, because you never learn who you lost.
Both are expensive. The first costs the replacement cycle above. The second costs the talent your business needed and could not identify.
What actually changes the math
There is no zero-risk hire. There is a measurable difference between gut-feel processes and evidence-based ones. Three concrete moves cut the risk on most SMB hires:
- Define the criteria before you see the first candidate. Not a job description. The 4 to 6 specific things this role requires, with weights and a rubric for each (what is strong evidence, what is weak). The interview scorecard template walks through how.
- Ask the same opening questions of every candidate for the same role. Depth adapts to each conversation, but the starting point is identical. Without this, you are comparing five conversations that happened in five parallel realities, and the decision falls on whoever you “remember best”, usually whoever spoke most fluently.
- Capture evidence during the conversation, not impression after it. Write down the candidate’s actual words against specific criteria. The decision gets made by reading evidence, not recalling vibes.
These three moves are the substance of what the literature calls the semi-structured interview, and they reduce both failure modes because they shift the decision from impression to evidence. The science behind them, and the studies above, is laid out on the methodology page. The full process for building and running that interview is in how to run a structured interview.
Frequently asked questions
How much does a bad hire actually cost?
The most rigorous public estimate is from the Center for American Progress (Boushey and Glynn, 2012), which reviewed 31 empirical case studies and found a median replacement cost of about 21% of annual salary for roles under 75,000 dollars a year. That measures replacing someone who leaves. A bad hire who stays, underperforms, and is eventually let go costs more. A transparent model for an 80,000 dollar role that ends at six months lands around 50,000 to 65,000 dollars.
Is it true that a bad hire costs 30% of annual salary, per the U.S. Department of Labor?
We could not trace that figure to any primary Department of Labor publication. It is repeated across HR blogs, usually cited as “U.S. Department of Labor, 2003” with no document or link, and a direct search of dol.gov returns nothing matching it. We treat it as uncited folklore and do not use it. The defensible number is the Center for American Progress 21% median.
What does the academic research say about the cost of turnover?
Peer-reviewed meta-analyses (Park and Shaw, 2013; Hancock et al., 2013) confirm that higher turnover is associated with worse organizational performance, but they report correlations, not dollar costs. No randomized study establishes a precise causal dollar cost of a bad hire. The dollar figures in circulation come from industry research and think-tank synthesis, which we cite and label as such.
Why do bad hires keep happening?
The structural cause is unstructured interviews scored from impression rather than evidence. McDaniel et al. (1994) found structured interviews substantially more predictive than unstructured ones (mean validity around .44 vs .33, reaching .63 in the strongest panel-consensus conditions), and Schmidt and Hunter (1998) reported .51 vs .38. Without pre-defined criteria and a rubric, the decision falls on whoever the interviewer remembers most favorably.
Does a bad hire cost more in a small business than in a large one?
Yes, proportionally. A large organization dilutes one bad hire across hundreds of employees. In a team of 5 to 15 people, one poor performer takes a meaningful fraction of the payroll, absorbs manager bandwidth, and affects the whole team. The formula is the same; the operational and cultural impact is amplified.
How Recrutador prevents this cost
The math above is structural, not accidental. Every source of error described (unstructured interview, impression-based decision, no written criteria before the role opens) is preventable, but only if the discipline exists and survives every hire. Most SMBs cannot maintain that discipline manually, which is why the same mistake repeats role after role.
Recrutador is a Hiring Intelligence Platform with five phases: the Strategist (a chat-first AI consultant) defines the role’s Evaluation Criteria (with weights, rubrics, and a probe library) before any candidate enters the funnel; the system generates the job description from those criteria; resumes are triaged with per-criterion coverage so you enter each interview knowing which gaps to probe; the live HUD runs a semi-structured interview during the call, surfacing the next probe one action at a time (every candidate starts from the same probe library, depth adapts per answer); and the Hiring Memo is generated automatically with quoted evidence per criterion. The same engine works for any role, at any seniority. There is no free tier; pricing is pay-per-use.
Want to see it on your next hire? Talk to the team and we run your first interview with you.
References
Footnotes
-
Boushey, H., and Glynn, S. J. (2012). There Are Significant Business Costs to Replacing Employees. Center for American Progress. Review of 11 empirical papers / 31 case studies; median replacement cost about 21% of annual salary for roles under 75,000 dollars. Article PDF ↩
-
Recruitment and Employment Confederation (REC) with Indeed (2017). Perfect Match: Making the Right Hire and the Cost of Getting It Wrong. Modeled cost of a poor mid-level hire (42,000 pound salary) exceeding 132,000 pounds. Modeled estimate, relies on a contested 40% failure-rate input. Press release ↩
-
Park, T. Y., and Shaw, J. D. (2013). Turnover Rates and Organizational Performance: A Meta-Analysis. Journal of Applied Psychology, 98(2), 268-309. DOI ↩
-
Hancock, J. I., Allen, D. G., Bosco, F. A., McDaniel, K. R., and Pierce, C. A. (2013). Meta-Analytic Review of Employee Turnover as a Predictor of Firm Performance. Journal of Management, 39(3), 573-603. DOI ↩
-
McDaniel, M. A., Whetzel, D. L., Schmidt, F. L., and Maurer, S. D. (1994). The Validity of Employment Interviews: A Comprehensive Review and Meta-Analysis. Journal of Applied Psychology, 79(4), 599-616. PDF ↩
-
Schmidt, F. L., and Hunter, J. E. (1998). The Validity and Utility of Selection Methods in Personnel Psychology: 85 Years of Research Findings. Psychological Bulletin, 124(2), 262-274. DOI ↩
-
Wingate, T. G., Bourdage, J. S., and Steel, P. (2025). Evaluating Interview Criterion-Related Validity for Distinct Constructs: A Meta-Analysis. International Journal of Selection and Assessment. DOI ↩