What a bad hire costs, and where verification sits against it

Why there is no number at the top of this page

Search for the cost of a bad hire and you will find confident figures: a multiple of annual salary, a specific rupee amount, a percentage of revenue. Follow any of them back and the trail usually ends at another article quoting a third, or at a vendor's marketing page with no method attached.

Some of those figures may well be roughly right for some organisations. The problem is that none of them come with the conditions that would let you know whether yours is one of them: which roles, which sector, what was counted, and whether the person was replaced or the position closed.

They also get used in a specific and unhelpful way. A verification provider quotes a large number for a bad hire, places a small number for a check next to it, and the arithmetic appears to settle the question. It does not, because the two numbers are not connected: the check only helps if verification would have caught this particular problem, and most bad hires are not misrepresentation.

So this piece does the other thing. It gives you the components, and you supply the numbers, because you have them and we do not.

Six components

Take a role. Assume a person joined, it went wrong, and they left within the first several months. Here is what it cost, in the order the money leaves.

1. Cost of the original hire, repeated

Whatever you spent filling the position the first time, you are spending again: agency or job board fees, the internal recruiter's time, interview hours across the panel, and the assessment process. Most organisations know this figure per role because they budget it.

2. Unproductive salary

Salary and cost-to-company paid over the period between joining and exit, less whatever value was actually delivered. For a role with a long ramp-up this is the largest single line, because a person who leaves at month five in a role with a four-month ramp-up delivered almost nothing.

3. Manager, team and HR time

The most consistently underestimated component. It includes onboarding and training hours, the manager's time managing a problem, the team's time absorbing the work, and then the HR and legal time spent unwinding the situation. Count the hours and price them at loaded cost; the total is usually larger than people expect.

4. Onboarding and provisioning

Equipment, licences, access provisioning, training programmes, background verification itself, relocation if any. Some is recoverable, much is not.

5. Direct loss

Only present in some cases, and dominant when it is. Theft, fraud, data taken to a competitor, a customer lost, a regulatory penalty, damage to a client relationship, remediation. This is the component that turns a moderate cost into a serious one, and it is also the component most closely connected to whether the person was who they said they were.

6. Disruption

Delayed projects, a vacancy carried again, lost momentum in a team, and where the exit was messy, the effect on people who watched it happen. Harder to price. Worth naming even if you leave it at zero, so nobody mistakes the model for a complete account.

The honest part: what verification does not touch

Bad hires come in at least three kinds, and verification addresses one of them.

Any provider who implies that verification reduces your total bad-hire cost is quietly assuming all three rows. It reduces the third row, and it reduces part of the second.

That sounds like an argument against buying it, and it is not, for one reason: the third row is where the direct loss lives. A poor-fit hire costs you components one to four. A misrepresented hire in a role with real access is the one that produces component five, and component five is the one that is not bounded by a salary.

So compare against exposure, not against a statistic

The right comparison is not "cost of a check" against "cost of a bad hire". It is "cost of a check" against "what could a misrepresented person in this specific role actually reach".

Work it through for three roles and the shape becomes obvious.

• A payments operations analyst. Reaches money, systems and customer records. A person who is not who they claimed, in that seat, can produce a loss that is not related to their salary and a regulatory problem on top. The check is trivial against that.
• A field technician entering customers' homes. Reaches customers, premises and company property. The exposure is safety and reputation, and neither is bounded either. Address, identity and court record earn their place immediately.
• An internal analyst with no system access, no budget and no customer contact. The exposure is the cost of replacing them. Identity, employment and education are cheap and worth running; a credit check and a directorship check are spend against an exposure that is not there.

This is the same conclusion the scoping guide reaches from a different direction, and it is why a single company-wide package is usually the wrong trade: it prices every role as though it were the middle one.

Building the internal case

If you have to argue for verification spend, the argument that survives scrutiny has four parts and no industry statistics in it.

1. The six-component figure for one real bad hire in your organisation. Yours, not a published multiple.
2. The exposure map. Which roles can reach money, data, customers, premises, goods or a signature, and how many people you hire into them a year.
3. The check cost for those roles only. Not for the whole headcount, which is what makes most of these business cases lose.
4. An honest statement of what it does not cover. Including that poor fit is the bigger problem and this does not address it. A case that claims less is believed more, and it survives the first bad hire that verification could never have prevented.

One thing worth measuring afterwards

Most organisations never look at what their verification programme actually found. A year in, it is worth pulling out: how many cases produced a substantive finding, what those findings were, and which roles they were in.

That gives you two things no vendor can. It tells you whether the packages are pointed at the right roles, because if every substantive finding is coming from one job family and you are checking eleven, that is information. And it gives you a real internal number for the next budget conversation, which beats a model, including this one.