Most founders hire a CTO twelve months too early or eighteen months too late. Same reason every time. They were watching the wrong number.

The funding round. The engineer headcount. The ARR line. All lagging indicators. By the time any of them fire, inertia has made the decision for you, badly.

This is the long version of a view I’ve argued in 7 things founders don’t want to hear about tech leadership. With the numbers, and the actual signal to watch.

The two bad playbooks everyone repeats

Two dominant scripts founders inherit. Both cost real money.

The first is the VC script: hire your technical co-founder or CTO on day one. YC’s guidance reinforces this for pre-product teams, and for a two-person garage startup it makes sense. The problem is non-technical founders reading the same advice at Seed, with a working product and a lead engineer already shipping. They hire a CTO title on top of a team that doesn’t need one yet.

You get a senior person with nothing sized for them. Karl Hughes at amazingcto.com has documented the pattern: bored CTO, title inflation blocking future hires, misallocated equity, execution slowing because architecture debates now need a decision-maker who didn’t build the thing. The Index Ventures Rewarding Talent data pegs the European Series A/B CTO equity benchmark at 0.7 percent. Give that away twelve months early and you cannot get it back.

The second script is the bootstrapper reflex: wait until it breaks. Ship until something on fire forces the hire. Cheaper for longer. Then it costs you a quarter of dropped velocity, a senior engineer resignation, or a security incident that closes a customer deal. I’ve watched founders lose their best engineer in the exact month they finally opened the CTO search. The engineer had been doing the job unpaid for six months and decided they’d rather do it somewhere that recognised it.

Both scripts anchor to inputs the founder can count, instead of the one signal that predicts the need.

The signal that actually matters

Your senior technical person has stopped shipping code and started making architecture calls they’re not confident in.

Week to week, it looks like this. Monday in a vendor selection call for the data warehouse. Tuesday rewriting a proposal from a Big Consulting shop because the numbers don’t add up. Wednesday interviewing two engineers, badly, because nobody wrote a rubric. Thursday arguing with the CEO about whether to migrate off the monolith. Friday they push one small PR at 8pm because it’s the only real code they’ve written all week, and it’s the part of the job they still love.

They’re not complaining yet. Too senior for that. But the commit history shows the drop. The Slack DMs are asking a friend at a bigger company what to do about SOC 2. The calendar is full of decisions they don’t have training for.

That’s the signal. It doesn’t need a headcount or a round to fire. It fires when the load of ownership crosses the line of confidence for the most senior person you already have. Miss it by a month and you’re fine. Miss it by six and they’re out.

Why headcount and revenue are lagging indicators

The rough industry marker is 15 to 20 engineers. Roughly the point where you cannot run engineering as a peer group, and you need someone whose full-time job is the shape of the org, the architecture, and the bar. A useful marker. Not a rule.

What moves the number up or down:

  • Regulated domain. Fintech, health, defence. You need the CTO conversation at five engineers, not fifteen. Architecture and audit posture from day one determine whether you can sell at all.
  • Complex integration surface. Stitching six vendors together, each with its own failure mode, lands the load on one senior person early.
  • Deep specialist team. A staff ML engineer who’s already run a team absorbs some load naturally. You can push the CTO hire out.
  • Enterprise sales in play. Enterprise buyers ask for a name and a title on the security questionnaire. A real, non-vanity reason to hire earlier than you’d want.

The Ravio 2026 Compensation Trends report shows median Series A headcount has shrunk from 57 in 2020 to around 47 in 2025, with early-stage hiring rates at 27 percent, down from 49 percent two years prior. Series A companies are smaller and hire slower than they used to. The old “you need a CTO at Series A” rule was already stale. It’s getting staler. The role arrives at a smaller headcount now than three years ago, which makes the signal more useful than ever, because the numbers no longer help.

Five symptoms you’re already past the signal

Here’s the checklist I run on a first call. Any one means you’re at or past the signal. Two or more means you’re overdue.

  1. Deploys have slowed. Not because the code got harder. Because nobody’s confident enough to press the button on Friday. Ownership has diffused.
  2. Security and compliance debt is piling up. Pen test findings from six months ago are still open. Nobody owns SOC 2. The CEO is answering security questionnaires by pattern-matching.
  3. Hiring has stalled. Two senior engineering offers turned you down in a row. Or you can’t get to the second interview because there’s no engineering leader on the panel who can sell the technical vision.
  4. Roadmap arguments are unresolved. The same architecture debate has come up in three consecutive planning sessions and nobody’s ended it. The CEO has started making the call by default, and getting it wrong.
  5. The founder is writing production code at midnight. Not because they love it. Because the alternative is the feature slipping. If you were technical once and you’re back at the keyboard against your will, the org has told you what it needs.

What a CTO actually does at Seed to Series A

Founders think hiring a CTO means hiring someone to manage engineers. It doesn’t. At Seed to Series A, the job is four things.

Architecture calls. Decisions cheap to make now and expensive to unwind later. Data model. Multi-tenancy. Auth. Which parts of the stack you build and which you buy. The person owning these needs to have been wrong about them before, in a previous company, and remember what it cost.

The hiring bar. Not the process. The bar. The CTO’s actual first-quarter job is often just interviewing every engineering candidate and setting the floor. This is where the money is. SHRM’s numbers put a bad senior hire at 50 to 200 percent of annual salary, and executive bad hires at an average of $240k total cost. With Ravio’s median mid-level engineer salary at £69,600, one bad senior hire costs you more than a fractional CTO for a year.

Vendor decisions. Which observability stack. Which data warehouse. Which auth provider. These decisions compound. Getting the top five right in year one saves a six-figure migration in year three.

The board-level tech narrative. How you talk about the platform, the risk, and the moat to investors and enterprise customers. Founders who try to do this themselves either oversell and miss, or undersell and lose the room. A real CTO makes the technical story defensible.

Notice what’s not on the list. Line management. Sprint ceremonies. Standup facilitation. Those are engineering manager jobs. Hire a CTO to run standups and you’re paying triple for the wrong role.

Full-time, fractional, or interim: the honest choice

Once the role is due, the next question is what shape it takes.

Full-time. Right when you have 18 months of runway, an engineering org large enough to justify a leader’s whole week, and enterprise sales requiring a named executive. UK Series A CTO cash comp lands between £150k and £220k depending on stage, plus that 0.7 percent equity. You’re committing for years, and if you get it wrong, the CB Insights Why Startups Fail data has this squarely under its 23 percent “not the right team” bucket. The most expensive bad hire in the company.

Fractional. Right when the load has crossed the signal but the org isn’t big enough for a full-time exec. Two or three days a week of a senior operator, permanently. Works when the founder wants a long-term partner but the cost of full-time doesn’t fit yet. UK 2026 pricing lands in a predictable band. I’ve written the honest version in this pricing guide.

Interim. Right when there’s a specific window. Pre-raise. Pre-launch. Post-departure. Someone senior in the seat, full-time, for three to six months, running the play and handing off. Different job to fractional. I’ve written the difference at length in fractional vs interim.

If your instinct is “we probably need a CTO but I don’t want to commit to a full-time hire until we know what the job actually is,” fractional is the honest answer. That’s the shape I run most often. Here’s what a fractional engagement looks like from day one.

The 30-day test before you commit

You don’t have to bet the company on a full-time hire to find out whether the role is real. Run a 30-day test.

Bring in a fractional or interim CTO for one month. Three deliverables. First, a written architecture assessment of what’s in place and what’s fragile. Second, a hiring plan for the next two roles with a rubric and a scorecard. Third, a security and vendor posture review, honest, in plain English, that you could hand to an enterprise buyer without wincing.

Thirty days in, you know three things you didn’t. Whether the role earns its cost. Whether the person is the right person, if you’re considering hiring them permanently. And whether the load you thought needed a CTO was actually a CTO problem, or an engineering manager problem, or a hiring problem, or a founder-focus problem.

Cheapest way I know to answer a question that costs six figures to get wrong. If you’ve hit the signal, book a 30-minute call. No deck, no pitch. You describe the situation, I tell you honestly whether you need the role and which shape fits.

If you’re still not sure

Three questions. Answer them this week, in writing, before you talk to anyone.

  1. In the last four weeks, how many hours has your most senior technical person spent writing code versus making decisions they’re not confident in? If code is under half, you’re at the signal.
  2. If your top engineer resigned tomorrow, how many architecture decisions would freeze? More than three, you have a bus-factor problem that a CTO solves.
  3. What’s the biggest technical decision you’ve made this quarter, and how sure are you it was right? If the honest answer is “not sure,” you’re already paying the cost of not having the role.

If any of those land uncomfortably, the signal has fired. Run the five-minute scorecard before the next round, the next hire, or the next planning session gives you a reason to wait.

Share: