The uncomfortable part is simple to state. You are about to make one of the two most consequential hires of the company, into the one domain where you cannot assess the work. Plenty of advice exists for this situation and most of it is a list of questions to memorise, which does not survive contact with a candidate who answers confidently and wrongly.
There is a better structure. It starts by not doing it alone.
First, decide which CTO you actually need
The title covers at least three different jobs. Hiring the wrong one is the most common and most expensive mistake in this process, because the mismatch does not show for six months.
| Kind | What they are good at | Right when |
|---|---|---|
| The builder | Writes code daily, ships fast, hires the first three engineers | You are pre-product-market-fit with under ten people. This is almost certainly you |
| The team builder | Recruits and runs 5–25 engineers, sets process, still reads code | You have traction and are scaling the team this year |
| The executive | Organisational design, budgets, board-level technology strategy | You are past Series B. Hiring this person early is a costly mismatch in both directions |
Write down which one you need before you write the job description. If your answer is "the first one, but I want someone who can grow into the second", say that out loud in the process — good candidates will tell you honestly whether that is them.
Borrow the judgement you do not have
This is the entire answer to "I cannot evaluate them". Find one senior engineer you trust who has no stake in the outcome — an angel investor with a technical background, a friend who is a staff engineer somewhere serious, a paid advisor for a handful of hours. Put them on the panel.
Pay for it if you have to. A few hours of a good engineer is trivially cheap next to a year of the wrong technical leader, and the difference in the quality of your shortlist is not subtle. You are not outsourcing the decision — the decision is yours. You are outsourcing exactly the part you cannot do.
Where these people actually are
Almost nobody good finds this role through a job board. The realistic channels, roughly in order of how well they work:
- Your investors. Every fund keeps a list of technical people they would back. Asking is free and they want this to go well.
- Engineers who already respect your space. People who have built in your market and understand why the problem is hard. They are findable by their writing and their talks.
- Second-in-command at a company one stage ahead. The strongest under-used pool: the person who is the most senior engineer somewhere but will never be CTO there.
- Your own senior hires, later. Sometimes the right answer is to hire two strong engineers now and promote in a year, once you can actually judge.
- Specialist recruiters. Expensive and occasionally worth it, but they cannot tell a great technical leader from a confident one either. They widen the funnel, not the judgement.
What you can assess perfectly well on your own
Your technical panellist handles depth. These signals are yours, and they are not the lesser half.
- How they explain something you do not understand. Ask about a hard decision from their last role. A strong candidate will pitch the explanation at you, check whether it landed, and use analogies instead of vocabulary. Someone who cannot do this with you will not do it with your board, your customers, or your sales team.
- What they ask you. Weak candidates ask about the tech stack. Strong ones ask who your customers are, what you have already tried, why the last approach failed, and how decisions currently get made.
- How they describe disagreements. Ask about a time they were overruled. Listen for whether they can state the other position fairly. A technical leader who cannot articulate the opposing case will not build a team that disagrees with them safely.
- How they talk about the people they hired. Specific names, specific growth, specific mistakes they made as a manager. Vagueness here is the reliable predictor of whether anyone will still be there in a year.
- Whether they tell you something you did not want to hear. During the process. About your product, your timeline, or your plan. The ones who do are the ones who will do it when it matters.
Run a real assessment, not a whiteboard
The best final stage for this role is paid and real: two or three days, properly compensated, reviewing your actual product and delivering a written assessment — what they found, what they would do in the first ninety days, in what order, and why.
You get three things at once. Your technical panellist can judge the depth of the findings. You can judge the writing, the prioritisation and the honesty. And you both find out what working together is actually like, which no interview reveals.
Red flags you can spot without technical skill
- Certainty before information. Anyone who tells you what your architecture should be before reading it is performing, not assessing.
- Everything previous was terrible. The engineer who inherited nothing but disasters at three companies is the common factor.
- Cannot name a mistake. Not modesty — a genuine inability to describe one is a serious signal for a role built on judgement under uncertainty.
- Rewrite in the first conversation. See Rewrite or Repair?. It is the instinct; it is not yet a diagnosis.
- Talks about technology more than customers. For the first CTO of a pre-seed company, this ordering is the whole job.
What to do while the search runs
Good searches take three to six months and you should not rush this one. The mistake is treating that time as a hold — the product keeps accumulating the cost described in Why Fast-Built Products Get Expensive Right After Funding while you interview.
The productive version is to spend the search period making the product legible: written down, releasable by more than one person, checked where it touches money. That work pays off twice. It shortens the ramp-up for whoever you hire, and it gives every candidate something concrete to react to — which, conveniently, is also the best assessment you will get of them.
