Key person risk in due diligence is the risk that the product stalls, or the service fails, when one particular person leaves. In software businesses that person is usually the CTO, a technical co-founder or a long-serving lead engineer. You test it by checking how concentrated the commit history is, who holds admin access and credentials, whether documentation has been used by anyone other than its author, and who handles production incidents. Then you price it and fix it through retention, deal structure and the first 100 days.
It's the finding I'd look for first on a lower mid-market software deal. The code can be ugly and the business still fine. If the only person who understands the billing system is about to receive a large cheque and has no reason to stay, the business is not fine. The wider tech diligence areas are in the technology due diligence checklist, the numbered data-room requests are in our technical due diligence checklist, and this piece goes deep on one of them.
Brilliance has little to do with key person risk. The test is whether anyone else could do their job by Friday.
What is key person risk in technical due diligence?
Brilliance has little to do with key person risk. The test is whether anyone else could do their job by Friday.
Software engineers have a name for this: the bus factor, or truck factor, meaning the number of people who would have to disappear before a project stalls. It's more common than most deal teams assume. A study of 133 popular GitHub projects by Avelino, Passos, Hora and Valente, presented at ICPC 2016, found 65% had a truck factor of two or less. Those are well-known open source projects with many contributors. A private company with eight engineers and a founder who wrote version one is usually more concentrated, not less.
The risk is sharpest at exactly the moment you are buying. Daniel Kim's research at MIT, using US Census Bureau data on around 4,000 tech startup acquisitions, found that 33% of acquired workers left within the first year, against 12% of comparable regular hires. That's US data on acquisitions by larger firms, not UK buyouts specifically, but the mechanism is the same. A transaction gives people both the money and the moment to leave.
How to tell whether the product survives the CTO leaving
There are four tests. Each one alone can mislead. Together they are reliable.
Commit history concentration
Ask for read-only access to the repositories and look at who has changed what, over the last 12 to 24 months, broken down by the parts of the system that matter: billing, permissions, the core data model, integrations, infrastructure as code.
The headline measure is the share of commits or changed lines per author, but the useful view is by area. A CTO with 30% of commits overall may still be the only person who has touched the billing service in two years. Look at recency too. Knowledge decays, and someone who wrote the payments code in 2021 but hasn't touched it since may know it less well than the engineer who fixed it last month.
Commit data has known traps. Squash merges credit one person for a team's work. Pair programming shows up under one name. A lead who reviews everything may write few commits but know everything. So treat the numbers as a map of where to ask questions, not as the answer.
Access and credentials
Ask for a list of who has admin rights to the production cloud account, the code hosting organisation, the domain registrar and DNS, the app store developer accounts, the payment provider, the email and identity provider, and any secrets manager or password vault. Ask which accounts are company-owned and which are tied to a personal email.
This is the cheapest test and often the most revealing. The CTO who registered the domain on a personal card in 2016 and is the only owner of the cloud root account is a key person whatever the commit history says. Fixing it is cheap. Not knowing about it is expensive.
Documentation that someone else has used
Every target will say it has documentation. Ask instead for the onboarding material the most recent hire used in their first fortnight, and ask that hire, in their interview, what was missing. Ask for the runbook for restoring the production database from backup, and when it was last followed by someone other than its author.
Good looks like documentation a newer engineer can describe using. Worrying is a wiki last edited by the CTO two years ago.
On-call and incident handling
Ask for the incident log and on-call rota for the last 12 months, and look at who resolved each serious incident. If the same name appears on every one, including the ones at 3am on a Sunday, that person is holding the service up.
A healthy pattern is a rota of at least three people, with serious incidents resolved by different engineers. A team where the CTO is "always on call" has a bus factor of one in operations, regardless of how the code is shared.
The interview questions that test the same ground from another direction, including the ones to ask engineers without the CTO in the room, are in technical due diligence questions.
Evidence to request for a key person risk review
| Evidence | What it shows | Red flag |
|---|---|---|
| Read-only repository access, 12 to 24 months of history | Who changes which parts of the system, and how recently | One author behind most changes to billing, permissions or the core data model |
| Admin and owner list for cloud, code hosting, domain, DNS, app stores, payments, identity | Who could lock the business out, or be unable to get in | Single owner, personal email or personal card on any critical account |
| Onboarding documentation and runbooks, with last-used dates | Whether knowledge exists outside heads | Docs written by and only used by the key person |
| Incident log and on-call rota, 12 months | Who keeps the service running | Same person resolves every serious incident |
| Org chart with tenure, contractor list, two years of leavers | Depth of bench and attrition | Senior departures, core code written by an unretained contractor |
| Employment contracts and IP assignments for key people | Notice, restrictive covenants and ownership of what they wrote | Founder code written before incorporation with no assignment |
| Current and proposed retention arrangements | Whether the key person has a reason to stay | Large cash-out at completion, no rollover, no retention |
The IP point needs a moment. Under section 11 of the Copyright, Designs and Patents Act 1988, copyright in work an employee creates in the course of employment belongs to the employer. Code a founder wrote before the company was incorporated, or as a contractor before joining, does not, unless it has been assigned in writing. When the key person is also the original author, check both.
How to price key person risk
Price it as the cost of reducing the dependency to something acceptable, plus an allowance for the disruption if the person leaves before that work is done.
The first part is usually a hire, an overlap period and some engineering time. Take an illustrative target where the CTO holds sole knowledge of the core platform. Assume a senior engineer at a loaded cost of £110,000 a year, a recruitment fee of £20,000, six months of overlap before the new hire can run releases alone, and two months of a second engineer's time on documentation and access transfer at £85,000 a year loaded. That's £20k plus £55k plus £14k, about £89k, before any delay to the roadmap. Those figures are assumptions for the example, not market data. Use the target's own salary bands.
The second part is harder. If the CTO left in month two, how long would the product roadmap stall, and what would that do to the revenue plan? A delay of six months to a product line worth £1m of year-two revenue is a number the model can use.
Kolonell's published guide to London technical due diligence puts the remediation cost of a key developer dependency at €40,000 to €90,000, covered by hiring and handover. That's one vendor's estimate rather than market data, but it is in the same range as the illustration above.
How to mitigate key person risk in the deal
The mitigations fall into three groups: keeping the person, structuring the deal, and reducing the dependency. You almost always need some of all three.
Retention
The simplest tool is a reason to stay. For a founder CTO, rolled-over equity in the new structure, with good and bad leaver provisions, does more than a cash bonus, because it ties them to the exit rather than to a date. For a lead engineer who isn't a shareholder, a retention bonus paid in instalments over 12 to 24 months, or a grant in the management incentive plan, is the usual route.
Check the person's existing contract too: notice period, restrictive covenants and garden leave. A three-month notice period with no non-solicitation clause gives you less time than the 100-day plan assumes.
Retention only works if the person wants to stay. Ask them, in the interview, what they want to do next. A founder who is visibly ready to stop is a founder whose retention package will buy you a notice period and little else.
Deal structure
Where the dependency is serious, put part of the consideration behind it. Deferred consideration or a retention held in escrow, released when handover milestones are met, ties the money to the outcome you need: access transferred to company accounts, a second engineer leading releases, documentation tested by a new joiner.
Be careful with earn-outs. An earn-out tied to revenue or EBITDA retains a founder CEO reasonably well. It does little for a CTO whose work affects revenue only indirectly, and it can create the wrong incentive, where the CTO keeps control of knowledge because control protects the earn-out. If a CTO is part of the earn-out, add knowledge-transfer milestones to it.
Make the cheap fixes conditions of completion. Transfer of admin access to company-owned accounts and signed IP assignments from the founder and any early contractors should happen before money moves, not in the first 100 days.
Source code escrow, often mentioned in this context, is a tool for software customers protecting themselves against a supplier failing. It doesn't help a buyer who is acquiring the code. It's relevant only if the target itself depends on a critical supplier's software, in which case check whether an escrow agreement exists and whether a change of control affects it.
The first 100 days
The rest goes in the 100-day plan, with owners and dates:
- Move every critical account to company ownership with at least two administrators and multi-factor authentication.
- Put a second engineer on every critical system, with them leading at least one release and one incident response.
- Hire the senior engineer or engineering manager the structure needs, starting the search before completion if you can.
- Run a documentation sprint, then test it by having someone new restore a backup or deploy a change from the runbook alone.
- Set up an on-call rota of at least three people.
Measure the result the same way you measured the risk. If, six months after completion, the commit history and the incident log still show one name, the dependency has not moved, whatever the plan says.
Key person risk beyond the CTO
The same tests apply to people the org chart doesn't flag. A long-serving contractor who wrote the integration layer. The one person who understands the data warehouse that produces the monthly board pack. An outside agency with the only copy of the mobile app signing keys. In my experience, the dependency that causes trouble after completion is often one level below the CTO, because nobody thought to check.
In AI-heavy targets there's a newer version: the one engineer who built the prompts, evaluation sets and model integrations behind the product's AI features. That knowledge is rarely in the code alone, and is worth checking as part of AI due diligence.
Key person risk is one of the areas we test hardest in technical due diligence, because it's common, it's cheap to find and it's expensive to discover after the money has moved. How it is presented in the final deliverable is covered in what a technical due diligence report should contain, and the technical due diligence page sets out how we scope the work.
Questions
What is the bus factor in software?
The bus factor, also called the truck factor, is the smallest number of people who would have to leave a project before it stalls because nobody left understands key parts of the code. A bus factor of one means a single departure could stop development or put the running service at risk. In due diligence it is estimated from commit history, access rights, documentation and interviews.
Is key person insurance worth it for a CTO?
Key person insurance pays the business a sum if a named person dies or, under some policies, becomes critically ill. It helps with the cost of a disruption but does nothing about resignation, which is the far more common way a CTO leaves after a deal. Treat it as a small supplement to retention and knowledge transfer, not a substitute.
Should a founder CTO roll over equity in a buyout?
Usually, yes, if the business depends on them. Rolled-over equity aligns the CTO with the exit, which is longer and more valuable than most retention bonuses. Combine it with good and bad leaver terms so that someone who leaves early gives up some value, and with a written handover plan so the dependency shrinks over the hold period rather than being preserved by the incentive.
How long should a key person handover period be after an acquisition?
For a CTO or lead engineer with sole knowledge of core systems, six to twelve months is a sensible range, depending on how concentrated the knowledge is and how quickly a successor can be hired. The handover should have milestones: access transferred, a second engineer leading releases, documentation tested by someone new, incidents handled without the key person. Time alone is not a handover.
Does key person risk apply to contractors and agencies?
Yes, and it is often worse. A long-serving contractor or an outside agency can hold as much knowledge as any employee, with no notice period, no retention package and sometimes no IP assignment. Check commit history by author, not only by employee list, and treat a contractor behind most of the core code as a key person for diligence purposes.