Skip to content

BlogCraft

The cost of a bad hire: the honest calculation

The “30% of first-year salary” has no verifiable source. For an IT services consultant, the cost of a bad hire is counted in days of lost margin.

9 min read

The cost of a bad hire is not calculated as a percentage of salary, and the figure people quote to do it, 30% of first-year pay, has no source anybody can open. For a consultant at an IT services firm, the cost depends almost entirely on a variable that percentage ignores: how the client reacts. In the worked example below, the same failure costs about €18,850 if the client waits for the replacement, and over €65,000 if it does not.

This topic follows on from the counter-offer statistic, with one difference that makes it more useful. There, the wrong number served as an argument in front of a candidate; here, it is used to decide how much to spend on avoiding the mistake, and a wrong order of magnitude skews every decision that depends on it.

Where does “30% of first-year salary” come from?

From an attribution to the US Department of Labor that is nowhere accompanied by the document it invokes. The sentence almost always circulates in the same form: the average cost of a bad hiring decision can equal 30% of the individual’s first-year potential earnings.

The oldest trace we found is not a departmental publication but the description of a book by Carol Hacker, The Costs of Bad Hiring Decisions & How to Avoid Them, whose second edition came out in 1998 from St Lucie Press. The publisher’s text already attributes the figure to the department, with no reference. An HR Exchange Network article published in 2014 presents it as the “most common figure” and dates it to 2003, which does not hold: an estimate from 2003 cannot appear in a book from 1998.

We did not find any departmental document containing this estimate, which does not prove that none ever existed. What is established is more modest, and it is enough: nobody among those repeating the figure cites it from the source, and the date it is given changes depending on where you read it. That is the behaviour we have already described for other numbers in this industry, the behaviour of a well-mannered rumour rather than of a measurement.

The only neighbouring figure with an identifiable source is a 2017 CareerBuilder survey, reported by the American Society of Employers: 74% of employers say they have hired the wrong person at some point, and the average reported loss comes to $14,900 per mistake. It is an average reported by US employers across every kind of role, and it says nothing about what a consultant placed at a client costs.

Why a percentage of salary is the wrong instrument for an IT services firm

Because in an IT services firm, what is lost when a hire fails is proportional not to the consultant’s salary but to the assignment’s daily rate and to how long it still had to run.

A percentage of salary assumes a cost that grows with pay: a failed junior would cost little, a failed senior a lot. Yet a junior placed on the most important account in your portfolio can cost you the renewal of three other assignments, while a senior who fails internally, in a role with no client attached, costs mainly time. The base of the calculation is the wrong one, and no coefficient corrects it.

The percentage has a second, quieter flaw: it gives a single number where reality gives a very wide range, governed by a decision that is not yours to make. A tool that answers €13,500 whatever the client does helps you weigh no prevention spending at all, because it does not say where the risk sits.

What does a bad hire cost over and above a good one?

That is the only honest question, and it immediately removes part of what the usual calculations add up. The cost of a failure is measured against the scenario in which the hire would have succeeded.

The initial sourcing, for instance, is not part of it. The job ads, the referral bonus and the time spent in interviews would have been spent in exactly the same way for a consultant who worked out, and counting them amounts to charging the failure for what came before it. The second sourcing, on the other hand, is entirely due to the failure, as is the second onboarding.

The same reasoning applies to the assignment days already worked: if they were billed, they are not a loss. What is lost is the days paid without being billed, the days given to the client by way of repair, and above all the margin the assignment would have produced while you look for a replacement. That last term is where the order of magnitude is decided.

The honest calculation, line by line, on a fictional consultant

What follows is an illustrative example, and every assumption is there to be replaced by yours. We take an experienced consultant hired at €45,000 gross a year and placed on a twelve-month assignment billed at €550 a day. For the fully loaded cost we use the French average monthly cost of a consultant on the bench, €5,480 a month, or €250 per working day, which leaves a margin of €300 per billed day. Six weeks in, the client asks for a replacement.

Line Assumption Cost
End of the probationary period two weeks’ notice, so 10 unbilled days €2,500
Commercial gesture 5 assignment days not billed, to settle things with the client €2,750
Second sourcing job ads, referral bonus, recruiting time €3,000
Second onboarding equipment, admin, business manager’s time €1,000
Margin lost during the replacement 45 days to replace, about 32 working days at €300 €9,600
Total if the client waits €18,850

The notice period, for its part, is set by French law: an employer ending the probationary period of an employee who has been there one to three months gives two weeks’ notice, and a manager on a permanent contract can be on probation for up to four months, renewable once where the industry agreement allows it. The 45 days are the median time-to-fill for an IT consultant, and they are what moves the result: between the first decile at 18 days and the last at 81, the same failure costs €13,150 to €26,650.

The usual percentage would give €13,500, which is roughly the cost of a failure at a firm that replaces as fast as the first decile of the market. It is a floor presented as an average.

A second scenario, with the same assumptions: the client does not wait and gives the rest of the assignment to a competitor. There were about two hundred working days left to bill, so €60,000 of margin, to which the end of probation and the commercial gesture are added, for a total of just over €65,000. Nothing in the consultant’s salary could have predicted that gap of one to three and a half between the two scenarios.

What is a damaged client relationship worth?

It is worth a probability rather than a number, and it is the one part of the calculation you will find in no study. A client who had to ask for a replacement almost never terminates the framework agreement; it consults you a little less quickly the next time, or opens the following request to two suppliers instead of one.

The only honest way to estimate it is to look at your own history. How many accounts have had a client-requested replacement over the last three years, and how many new requests did each one send you in the twelve months that followed, compared with the twelve months before? Multiplied by the average margin of an assignment on that account, that difference is the relationship cost. It will often be higher than the whole table above, and nobody calculates it, because it shows up in no tool.

The cost of waiting, by contrast, is already well measured. Days on which a consultant is paid without being billed are the same mechanism as bench time, and Cobalt has set out what those days really cost an IT services firm and why they are underestimated. The difference is that a bad hire produces them twice: on the consultant who leaves, and on the assignment that waits.

The bad hire who does not leave

The most expensive case is the one no calculation captures, because it triggers no event. The consultant stays, the client puts up with it, the assignment is billed, and nothing gets written down anywhere.

It is paid for later and elsewhere: at renewal, which the client does not sign without anybody being able to say why, or on the next request from the same account, which goes to a competitor. It is also paid for by the consultant, badly placed on an assignment that does not suit them, who eventually leaves in turn. We noted about the bench time firms try to avoid at any cost that putting someone on an unsuitable assignment to keep them off the bench is paid for in a departure a few months later.

That is also why the first weeks matter so much. 33% of new starters leave within ninety days, as we wrote about onboarding, and that window falls entirely within the probationary period: it is the moment when a failure is cheapest to acknowledge, and the one at which people most often prefer to wait and see.

Where should prevention money go?

Where the calculation shows the cost forming, which is before signature and in the very first weeks, and not into a longer selection process.

Before signature, two checks reduce the probability of failure more than any additional test. The client brief, which decides the placement before the first CV, because a criterion misunderstood at the start produces a consultant who is competent and unsuited. And a reference check that verifies facts rather than collecting opinions, since the gap between what the candidate claimed and what is true is exactly what makes a client ask for a replacement in week six.

After signature, the lever is the date of detection. A failure acknowledged in week three rather than week ten costs fewer free days, less tension with the client, and lets the replacement start sooner. A fixed check-in with the consultant and another with the client, on a precise date, are worth more than an impression shared in passing at a steering committee.

The data this calculation needs, client-requested replacements, time to replace, requests received per account before and after, is already in your ATS and your CRM. It is the kind of figure we have Balt pull out at the moment the question comes up, rather than a conference percentage. That leaves the question that sets the heaviest term in the table: how many days do you need, today, to put a credible replacement in front of the client who asks for one?

Frequently asked questions

  • Did the US Department of Labor really say a bad hire costs 30% of salary?

    None of the pages that say so cites a document from the department, and we found none. The wording already appears in 1998 in the description of a Carol Hacker book on the cost of bad hires, and later articles date it to 2003, which is inconsistent. It should be treated as a rule of thumb with no established origin.

  • How much does a bad hire cost an IT services firm?

    There is no single honest figure, because the cost depends mainly on what the client does. On a fictional consultant billed at €550 a day, our calculation gives €13,150 to €26,650 if the client waits for a replacement, depending on how fast you replace, and over €65,000 if it gives the rest of the assignment to a competitor.

  • Should the cost of the first sourcing be counted in the cost of a bad hire?

    No. It would have been spent in exactly the same way had the hire worked out, so it does not measure the failure. What counts is what the failure adds: the second sourcing, the second onboarding, the days paid without being billed, and the margin lost while you wait.

  • How do you reduce the cost of a bad hire in an IT services firm?

    By shortening the time to replace and by spotting the failure earlier, since those are the two terms that weigh most. A precise client brief, a reference check that verifies facts and a check-in with the consultant in the first weeks of the assignment do more than any calculation after the event.

Sources

  1. Carol Hacker, The Costs of Bad Hiring Decisions & How to Avoid Them, 2nd edition (St Lucie Press, 1998), publisher’s descriptionschweitzer-online.de
  2. HR Exchange Network, What’s the real cost of a bad hire? (April 2014)hrexchangenetwork.com
  3. American Society of Employers, The cost of a bad hire (December 2017, CareerBuilder survey)aseonline.org
  4. Service-public.gouv.fr, Probationary period for an employee on a permanent contract (in French)service-public.gouv.fr
  5. Cobalt, Intercontrat en ESN : arrêtons de faire semblant que ça ne coûte rien (March 2026, in French)cobalt-ia.com

Balt does this work

Read next

Give Balt its first job.

Thirty seconds to add it to Teams and get to work.

Try for free