Cost of a Bad Hire: 39.6% of Annual Wage on Average

The most traceable estimate of the cost of a bad hire is the cost of replacing the person: 39.6% of the role’s annual wage on average and 23.5% at the median, pooled from 31 case studies across 13 research articles [1]. At the US median graphic designer wage of $62,960 that is $24,932 on average and $14,796 at the median [2].

One scope note belongs in front of every figure on this page. No published study prices a bad hire as its own category.

What the research measures is turnover: separating from somebody and replacing them. A bad hire is one reason turnover happens, so these numbers are a floor under the real total.

What they don't carry is everything the wrong person cost you while they held the seat.

From here on, this page traces the six figures you'll meet on this question back to the document behind each one, and prints what that document actually says.

Then it prices one failed design hire at the federal median wage. Where a famous number turns out to have no traceable origin, you'll see that said plainly.

What is the cost of a bad hire?

Put it at 23.5% to 39.6% of the role’s annual wage. Those are the median and the average of 31 pooled case studies on the cost of replacing a worker, and individual positions in that pool ran from 2% to almost 150% of annual wage [1].

Two economists at the Washington Center for Equitable Growth built that pool from 13 research articles published between 2000 and 2020, covering healthcare, education, hospitality, finance, retail, transportation and manufacturing jobs [1].

They also published the per-position table behind the averages. That's why theirs is the one estimate on this question you can check line by line.

Every other figure you meet here measures something different, and the three get blurred together constantly.

Cost per hire is recruiting spend to fill a seat. Cost of turnover is what it takes to replace somebody who's gone. Cost of a bad hire is that second one plus the damage done before they went, and no study prices the damage.

So the table below does the separating. It takes the six numbers in circulation and says what each one actually counts, who it counted, and when.

Only the first row is a pooled study with a published method. Three of the rest are single-source estimates from named organisations, and two of those decline to show the study, the sample or the method behind the number they publish. The last two aren't findings at all, so keep them out of your own budget.

How much does it cost to replace an employee?

It costs about 23.5% of annual wage to replace an employee in the median position and 39.6% on average, with individual jobs running from 2% to almost 150% [1].

For workers earning less than the 2019 average annual wage of $53,490, replacement came to 19.3% of their annual wage [1]. Which band your own role sits in moves the figure more than any headline average does.

Those splits are what make that range usable. In the two sectors with at least five case studies each, turnover cost 32.7% of average annual wage in health services and 19.6% in hospitality [1].

Pay level moves the figure more than industry does. In the pool, the floor of about 2% is a registered nurse and the ceiling of 148% is a lower-level executive, both from the same 2006 study of New Jersey employers [1].

That's the pattern to carry into your own numbers. The higher the salary and the more specific the role, the larger the share of that salary you lose replacing it.

Two other sources get quoted as often and neither one shows its work.

Gallup states that replacing an individual employee can cost one-half to two times their annual salary, and calls that conservative. Its own page names no study, no sample and no method [3].

Work Institute put it at 33% of base pay in its 2018 Retention Report, and the report credits that ratio to its own prior-year report, with no derivation shown [6].

One more separation belongs here, because it's the one that gets lost most often. Replacement cost is not cost per hire.

SHRM measured average cost per hire at $4,129 from 2,048 of its members on fiscal year 2015 data, and that figure covers the recruiting spend to fill a seat and nothing else [4].

So add it on top of a replacement estimate.

Use both ratios as cross-checks on your own figure. They bracket the pooled figures from either side, which is some comfort, and neither one can be audited the way a 31-row table can.

Does the US Department of Labor say a bad hire costs 30% of first-year earnings?

We went looking for a Department of Labor document that says it and couldn't find one. Every page repeating the claim traces back to a flat assertion with no citation of any kind [7].

So treat 30% of first-year earnings as unsourced, and keep it out of your own budget until somebody produces the document.

Here's the chain, because showing it is the whole proof. Google's AI Overview for this query gave the Department of Labor line as fact on 27 October 2026, and credited SocialTalent.

SocialTalent writes that “according to the U.S. Department of Labor, these missteps can cost up to 30% of the employee’s first-year earnings” and hyperlinks the words U.S. Department of Labor to an article on Business.com [7].

That article repeats the claim in almost the same words and links to nothing. No report name, no document number, no dol.gov address anywhere on the page [7].

A search of dol.gov for the phrase returns no such document either.

The $240,000 figure has a milder version of the same problem. SHRM's own article on the cost of a bad hire carries no SHRM-generated number at all.

It relays an estimate of “as much as $240,000” to recruit, hire and onboard a new employee, attributed to the chief executive of a London employer-branding agency with no method published [5].

That same article quotes a 2012 CareerBuilder survey in which 43% of employers said they made a bad hire because they felt they had to hire somebody quickly, and 22% said they lacked the skills to interview well [5].

Two studies belong in this trace and are missing from it. We'd rather tell you that than quote them at second hand.

The 2012 Center for American Progress brief by Heather Boushey and Sarah Jane Glynn is real and named: the Equitable Growth brief states that its own research follows on from it [1]. So is the 2012 CareerBuilder survey.

Neither primary page could be accessed, so no figure from either one is used above, apart from the two percentages SHRM quotes in its own article.

Look at the pattern in that table, because it is the finding. One source with a published method is the one almost nobody quotes, and the figure quoted most has no document behind it at all.

What does a bad design hire cost at the US median wage?

At the federal median graphic designer wage of $62,960 a year, those pooled ratios put one failed design hire between $14,796 and $94,440, with $24,932 as the average case [1][2].

Nobody publishes that multiplication, so here it is with every assumption in view, ready for you to redo at your own wage.

One line does all of it. Cost equals annual wage times the published ratio.

The wage is fixed at $62,960, the national median for graphic designers in May 2025 [2]. Every ratio comes from a source quoted above.

Results are rounded to the nearest dollar, so 62,960 x 0.235 is $14,795.60 and prints here as $14,796.

Three assumptions carry the whole calculation, and each one is a choice you can disagree with.

• The wage base is the national median, not the mean. A first design hire is usually made near the middle of the market, and a mean gets pulled up by senior specialists. The federal range runs from under $39,520 to more than $104,910 [2], so scale the result to the offer you'd actually make.
• Every ratio is cross-occupational. The 31-study pool holds healthcare, education, hospitality, finance, retail, transportation and manufacturing positions, with no design role among them [1]. So the model assumes a designer costs the same share of salary to replace as the pooled average of unrelated jobs, and there's no design-specific study to put in its place.
• The figure counts replacement only. It doesn't count the revenue, the rework or the client damage done while the wrong person held the seat, which is why every number here is a floor.

One more note on the wage base. The full distribution behind that median sits in our graphic designer salary report, and what the same hire costs once employment costs go on top of salary is in how much a graphic designer costs.

Replacement figures below sit on top of that, not inside it.

Gallup comes at the same question from a different direction and lands higher. One-half to two times salary puts this hire at $31,480 to $125,920 [3].

Its floor sits above four of the five pooled results in the table, so the two sources don't corroborate each other so much as mark out how wide the honest uncertainty still is.

If you want one number to plan against, take $24,932. It's the average of the only pooled dataset on the question, applied to the wage employers actually pay, and it's almost certainly low.

Where this model breaks

Seven limits, and all of them are real. If one of them matters to your situation, the figure above is the wrong figure for you.

• No study in the pool is about designers. Not one of the 31 case studies behind these ratios covers a design, a creative or a graphic role, and the brief lists every position it pooled [1]. The share of salary is borrowed across occupations, and that's the single biggest weakness here.
• Ratio and wage come from different decades. The case studies were published between 2000 and 2020 [1] and the wage is May 2025 [2]. A share of salary travels across years better than a dollar figure does, which is why we apply the ratio to a current wage and leave the old dollars where they were, and it still isn't the same year.
• The 19.3% band is deliberately the wrong band. It applies to workers earning less than the 2019 average wage of $53,490 [1], and a $62,960 designer sits above that line. It's in the table as a floor check, not as the right answer.
• The ceiling is one data point, and it is rounded. The $94,440 prices the brief's own published ceiling, almost 150% of annual wage [1]. Behind that range, the single highest row is 148%, a lower-level executive in a 2006 study, which at this wage comes to $93,181, because 62,960 x 1.48 is $93,180.80 [1]. Either figure marks the top of the published range. Neither one describes a typical bad design hire.
• Turnover cost is not bad-hire cost. Everything above prices separation and replacement. None of it prices the work shipped wrong, the deadline missed or the client who went quiet, so treat the total as a floor.
• The wage covers employees only. The federal survey excludes self-employed workers [2], so the median is what employers pay on payroll. That's the right scope for a hiring decision and the wrong one if you're comparing freelance rates.
• Two sources we wanted are absent. Both the Center for American Progress brief of 2012 and the CareerBuilder survey of 2012 could not be accessed. Their figures are missing rather than estimated, and we didn't copy them out of somebody else’s summary.

Hire a designer or use a design subscription?

Hire when the design work fills a full week all year. Subscribe when it comes in bursts and you want to be able to stop.

Those figures only settle the question if you put both options on the same axis, and that takes two divisions.

At the median wage, one designer costs $5,247 a month in salary before any employment cost goes on top, which is 62,960 divided by 12 [2].

A hire you end at month six has cost $31,480 in salary, plus the replacement figure from the table above. That's about $46,276 at the pooled median, and more if the role sits higher in the band.

On the other side, what you can lose before you can stop is one month. Cluma runs a 7-day trial, the first month is paid in full, and after that you can pause or cancel anytime with no long-term contract and no cancellation fee [8].

For uCreate, the everyday graphic design module, that first month is $3,495. A pause stops billing immediately and banks the unused days of the month you already paid for [8].

Now the fair part, because that table flatters the second row if you read only the money column.

A year of the module is $41,940 against $62,960 of median salary, and the salary figure is still before employment costs. So the real gap in your budget is wider than those two annual numbers make it look.

One module is also one request in progress at a time, which isn't the same thing as a full week of dedicated capacity [8]. Cluma’s modules run $2,995 to $4,995 a month, so a broader brief than everyday graphic design costs more than the row above [8].

When the hire wins: your design work genuinely fills a full week for twelve months, you need somebody with the product in their head rather than in a brief, you want them in the standups and in the room when the roadmap changes, and you want that knowledge to stay with the company.

On that workload a salaried designer is the better buy for you, and nothing in the table above changes it.

When the work comes in bursts, the salary runs every month anyway and the replacement risk sits with you.

For the subscription side provider by provider, our comparison of ten services goes through them one at a time, the return side of the same budget sits in our McKinsey design index write-up, and the rest of our sourced numbers are on the Cluma reports page.

If bursts are your pattern, a design subscription is the version of this budget you can switch off: $3,495 a month for uCreate, a 7-day trial in front of it, and a pause that stops the billing on the day you take it [8].

References

1. Washington Center for Equitable Growth, Kate Bahn and Carmen Sanchez Cumming, Improving U.S. labor standards and the quality of jobs to reduce the costs of employee turnover to U.S. companies, December 2020, read on 27.10.2026 including the methodological appendix and the full case-study table. Pooled research with a published method. Source of the 39.6% average, the 23.5% median, the 2% to almost 150% range, the 19.3% below-average-wage figure with its $53,490 threshold, the 32.7% and 19.6% sector splits, the 2.1% and 148% extreme rows, and the Center for American Progress lineage. Weaknesses worth knowing: the brief pools 31 case studies from 13 research articles published between 2000 and 2020, so the ratios are not from one year, none of the 31 positions is a design role, and the brief measures turnover rather than bad hires.
2. US Bureau of Labor Statistics, Occupational Outlook Handbook, Graphic Designers, May 2025 wage data, re-read 27.10.2026. Government source. Source of the $62,960 median annual wage and of the range from under $39,520 to more than $104,910. One weakness, and the page states it itself: the wage survey does not collect pay for self-employed workers, so the median is what employers paid on payroll.
3. Gallup, Shane McFeely and Ben Wigert, This Fixable Problem Costs U.S. Businesses $1 Trillion, 13 March 2019, read 27.10.2026. A named firm’s own published estimate. Source of the one-half to two times annual salary range. Its limitation is noted in the article above: the page gives the range and calls it conservative without naming a study, a sample or a method, so it reads as a house estimate and is used here only as an outside cross-check.
4. SHRM, SHRM Benchmarking Report: $4,129 Average Cost-per-Hire, 8 August 2016, read 27.10.2026. Member survey. Source of the $4,129 figure, taken from the Human Capital Benchmarking Report on a random sample of 2,048 SHRM members, reflecting fiscal year 2015. Weakness: it measures average cost per hire, which is recruiting spend to fill a seat, and it is routinely quoted as the cost of a bad hire, which it is not.
5. SHRM, Lisa Frye, The Cost of a Bad Hire Can Be Astronomical, 9 May 2017, read 27.10.2026. An article relaying third-party claims. Source of the $240,000 figure and of the two 2012 CareerBuilder percentages, 43% and 22%. Why it appears here at all is also its weakness: the article carries no SHRM-generated cost figure. That $240,000 is attributed to the chief executive of an employer-branding agency with no method published, and the survey percentages are CareerBuilder’s.
6. Work Institute, 2018 Retention Report, read 27.10.2026 from a Montana State University mirror of the original PDF, because the original on workinstitute.com could not be accessed and the current report sits behind a lead form. A named firm’s own published report. Source of the 33% of base pay figure, direct and indirect costs, drawn from a report built on more than 234,000 exit interviews. Weakness: the report credits the 33% ratio to its own prior-year report, Sears, Nelms and Mahan 2017, rather than deriving it, so it is a recurring house estimate and not a fresh calculation.
7. SocialTalent, What is the Cost of a Bad Hire?, and the Business.com article its Department of Labor link points to, both read 27.10.2026, and Google’s AI Overview for this query on the same day. Cited as evidence of a citation chain and never as the source of a number. These pages are how the 30%-of-first-year-earnings claim travels. That is the point of including them. Neither page cites a Department of Labor document, and a search of dol.gov for the phrase returns none.
8. Cluma, pricing, read 27.10.2026. Published by Cluma. Source of the $3,495 uCreate price, the $2,995 to $4,995 module range, the 7-day trial, the first month paid in full, the pause and cancellation terms and the delivery terms in the comparison table. Disclosure: Cluma publishes this report and is one of the two options in that table, on the same footing as the hire. No saving, return or advantage is claimed for it anywhere above, because we hold no data that would support one.

Three sources named in the brief for this report stayed out of it on purpose: the Center for American Progress brief of 2012, the CareerBuilder survey of December 2012 and the Work Institute’s current retention report. Each one is real, and each is either behind a hard 403 or behind a lead form for us today. Nothing from them is interpolated, estimated or copied out of somebody else’s roundup, and no figure in this report comes from a statistics roundup, a vendor summary or an AI answer.