The most expensive fractional CTO decision is not the rate. It is signing a six-month engagement with someone you have talked to twice, discovering in month three that the fit is wrong, and eating the notice period on the way out. There is a cheaper way to find out, and it is the thing good operators offer without being asked: a paid trial before either of you commits.
I run these both ways, as the person being hired and as the person advising founders who are hiring. Done right, a paid trial tells you more in three weeks than three more interviews ever will, and you keep something useful even if you walk.
Why a paid trial beats another interview
Interviews test how someone talks about work. A trial tests how they do it. The gap between the two is where most bad hires live. A candidate who sounds sharp on a call can still spend the first real week reorganizing your Jira board instead of touching the thing that is actually on fire. You only see that by watching them work on your actual problem.
The trial also flips the pressure. In an interview you are selling the company and they are selling themselves, and everyone is on best behavior. In a paid engagement they have to produce, and you get to see how they behave when the answer is genuinely hard: do they hedge, do they overpromise, do they ask the uncomfortable question about your codebase that the polite interview never surfaced.
Ask for it explicitly if it is not offered. A fractional CTO who resists any trial and only wants the long contract is telling you something. The confident ones would rather prove it than pitch it.
What a good trial looks like
The shape I use is a paid discovery sprint of two to six weeks, roughly one week of effort per month at the person's normal rate. You are not asking for free consulting and you should not accept it; unpaid trials attract people with nothing better to do, and they set a bad tone for a relationship built on judgment. Pay the real rate for a defined, small block.
The sprint should produce something you can hold. At the end you want a written technical assessment and a prioritized roadmap: what is on fire, what can wait, what the next 60 to 90 days should look like, and what they would defer and why. Crucially, that document should be useful to you even if you never work with this person again. If the deliverable only makes sense as a lead-in to their long contract, it was a sales exercise, not a trial.
A strong first 90 days follows a clear arc, and the trial is the front of it: assessment and a couple of quick wins first, then the strategy, then implementation. If the person cannot show you the assessment-and-quick-wins portion in a three-week paid block, the full engagement will not go better.
What to watch during the three weeks
Watch where their time goes in week one. A good operator spends it asking about your business, your team, and your current state before touching anything. Someone who arrives with strong opinions before they have read the code is pattern-matching your company to their last one, and that is how you get the expensive rewrite you did not need. This is the same instinct behind why the day rate is a bad way to compare people: the number tells you nothing about whether they will spend the week on the right problem.
Watch how they handle disagreement. Push back on one of their recommendations during the trial and see what happens. Do they defend it with evidence, fold immediately, or dig in on ego. You are hiring judgment, so test the judgment while it is cheap to test.
Watch the references in parallel. Talk to founders at a similar stage, not enterprise clients whose problems look nothing like yours. The skills that work at a 200-person company do not always transfer to a pre-seed seat, and the trial plus a same-stage reference together catch what either misses alone. When you are weighing the whole picture, it helps to have already worked out whether the spend is worth it at all.
The terms that make the trial clean
Put four things in writing before the sprint starts. The fee and what the paid block covers. The deliverable, named specifically, so there is no argument about what you are buying. A clear statement that the trial does not auto-convert to a longer contract, so nobody feels trapped. And ownership of the deliverable, so the assessment and roadmap are yours to keep and act on regardless of what you decide next.
Then hold a real decision at the end. The point of the trial is that walking away is a legitimate, low-cost outcome, not a failure. If you convert, you do it having actually seen the work. If you do not, you paid for a roadmap you can hand to the next person. Either way you avoided the six-month mistake, and you can see how a converted engagement is scoped so it does not creep into a full-time seat.
FAQ
How long should a fractional CTO trial run?
Two to six weeks, structured as roughly one week of effort per month at the person's normal rate. Long enough to produce a real assessment and a couple of quick wins, short enough that walking away costs you little.
Should the trial be free?
No. Pay the person's normal rate for a defined block. Free trials attract the wrong people and set a bad tone for a relationship built on paid judgment. You are buying a deliverable, not begging for a favor.
What should I get out of the trial?
A written technical assessment and a prioritized 60-to-90-day roadmap that is useful to you even if you never work with this person again. If the output only makes sense as a lead-in to their long contract, it was a sales pitch.
What if I decide not to continue after the trial?
That is a valid outcome and the whole reason to run one. You keep the assessment and roadmap, hand them to the next candidate or your team, and you have spent a few weeks of rate instead of eating a six-month notice period. Then book a call to talk through what the next person should own.