When should a founder hire a CTO?
Later than most founders expect, and earlier than they'd like. Five signals that the technical decisions have outgrown the person currently making them.
It reaches me as “we need a CTO”. What’s happened underneath is narrower: a series of technical decisions started costing real money and nobody senior was making them, and by the time it gets phrased as a hiring question the money has already gone.
A CTO is one answer, and the costliest. Five signals worth watching for.
1. You’re making architecture decisions by accident
The clearest early warning is a decision nobody recognises as one. Someone picks a database because it was on the last project, a queue goes in because a tutorial used one, and six months later you have an architecture nobody chose that now takes a project to change where it should have taken an afternoon.
The test: can anyone in your business name the three technical decisions that would be most expensive to reverse? If nobody can, those decisions have already been made by default.
2. Your engineers ask commercial questions and get no answer
Good engineers hit forks where the right choice depends on business context they don’t have. Build this properly or ship it in a week? Handle a thousand or a hundred thousand? Is it worse to be slow or to be wrong?
If nobody answers, engineers answer for themselves, and they answer as engineers: building for a scale you don’t have and a robustness you never asked for. Over-engineer it and you’ve wasted months; build it too thin and you’re paying for a rewrite a year later.
3. The estimates stopped being useful
Early on, estimates are wrong and it barely matters, until you make commercial promises against them: a customer date, an investor milestone, a hiring plan.
When estimates and reality keep diverging, the cause isn’t that the engineers estimate badly. The work is described too vaguely to size, or in so much detail that writing the specification becomes the job, and both of those get decided above the engineers.
4. You can’t answer a technical question from a customer, an auditor or an investor
A large customer sends a security questionnaire, an investor asks about technical risk in diligence, or a regulator asks where the data lives. If the honest answer is “I’d have to ask”, nobody owns the technical narrative, and nobody is managing the risk either.
5. You’re the bottleneck and you aren’t technical
The most common version, and the hardest to see from inside. Every technical decision routes through the founder, who lacks the background to make it quickly, so it either queues or gets made on instinct, and it might take somebody else to tell you.
What to do about it
Three options, in ascending order of commitment.
Buy a decision. If you have one large open question (which platform, build or buy, whether this architecture survives growth), buy an answer to that question, scoped and paid for and with an end date. It’s by far the cheapest thing on this list, and more than once it has shown that the other two can wait a year.
Buy ownership at a cadence. If the decisions are continuous, a fractional arrangement gets you someone accountable for technical direction without a permanent hire. £3,000 to £8,000 a month is what I see quoted and paid. That’s a working benchmark, and I haven’t surveyed the market. What a fractional CTO costs in the UK has the named sources and the dates they were checked.
For most businesses asking the question, the middle one is the right answer, and it stays right for longer than founders expect.
Hire permanently. When leading the people becomes a full-time job, you need someone in the building every day. That’s when £150,000-plus (the median in my pricing article is £160,500) and a three-to-six-month search become a reasonable plan.
I’ve watched businesses get this backwards: a full-time CTO hired to answer one question, then a leader’s salary paid for a team that didn’t exist yet.
What you’re really buying
At every level you’re buying someone who’s paid to be right about the big decisions and who carries it when they’re wrong. You’re paying for the judgement, and the stack on the CV will have changed by the time any of it matters.
I hold both jobs in my own company: I decide what gets built, and I’m the one who has to explain why a conversion rate moved. Where those two sit with two people who don’t talk to each other, the decisions get worse.