What does a fractional CTO actually do?

A fractional CTO is a chief technology officer you hire for a fraction of the time. Same seniority, same kind of decisions, a fraction of the hours — typically one to three days a week, on a retainer measured in months. That is the meaning; everything else is scope.

The distinction people miss is between an advisor and an executive. A consultant studies something and hands you a document. A fractional CTO sits in your leadership meetings, owns decisions, and is still there when one turns out badly — part-time in hours, not in accountability. Someone selling you a report and calling it a fractional CTO engagement is a consultant with a better title.

The hours go to four things.

Deciding what to build and what to buy. Which parts of your stack are worth owning, which should come off a vendor’s shelf, and which should not exist. Most of the money is won or lost here, and every vendor in the room has a view.

Judging vendors and their work. Proposals, quotes, contracts and the code that comes back: what a plausible timeline looks like, which line items are padding. On AI work this is usually the difference between a pilot and a bill — why AI projects fail is a longer note on it.

Running or repairing the engineering function. Hiring, team structure, how work ships and how quality gets checked — sometimes unpicking an agency relationship, sometimes telling you the team is fine and the roadmap is the problem.

Carrying technical credibility outward. The diligence call, the investor meeting, the security review — answering what your team cannot answer yet.

What it is not: writing your features. A fractional CTO in your codebase is an expensive senior engineer; hire an engineer instead.

A full-time CTO is an employee: all their hours, equity, a multi-year horizon. They build the engineering culture because they are there long enough for there to be one.

A fractional CTO is a part-time executive across several companies — ongoing, but bounded. Right when the decisions are serious but intermittent: the questions are big, they just do not arrive daily.

An interim CTO is full-time and temporary, holding the seat while you recruit. Interim is about coverage; fractional is about level.

The honest version of the comparison

The pitch is always “CTO-level judgment at a fraction of the cost”. True, and incomplete: you get a fraction of the judgment too, because judgment runs on context and context runs on hours. On strategic questions, one day a week is plenty. On questions that turn on who said what in which meeting, it is not.

The benefits are real, and narrower than the marketing suggests.

Seniority you could not otherwise hire. A twenty-year operator will take a few days a month with an interesting company long before a full-time role at one.

Speed on the decisions that matter. Someone who has read the same vendor proposal thirty times reaches a verdict in an afternoon that would cost your team six weeks and still end in a guess.

Neutrality — if you insist on it. An advisor with no build capacity to sell has no reason to recommend a build, but that only holds if it is in the contract. Ask whether they take commission, referral fees or revenue share from anyone they recommend. Nobody asks it.

An exit that breaks nothing. A good engagement ends. You keep the decisions and the documents, and you stop paying.

I do this work, so read the following knowing it costs me something to write down.

Split attention is not a hedge, it is the model. When your urgent thing lands on a day belonging to another client, it waits. If you need a decision-maker reachable within the hour, fractional is the wrong shape.

A retainer pays someone to still be necessary. This is the drawback the industry does not print. A monthly retainer creates a quiet incentive to remain useful next quarter — to keep the roadmap slightly unfinished, to be the one person who understands the architecture. Nobody has to be dishonest for that to work on them; it is what the arrangement rewards.

The handoff problem. If the reasoning lives in the advisor’s head rather than your documents, their departure takes your technical strategy with it, and six months later your team is running a plan nobody can explain.

Authority without consequences. Recommendations land long after the engagement closes. Ask anyone you are considering about a call they got wrong and what it cost the client; the ones who cannot are telling you something.

Five situations where a continuing arrangement beats a one-off opinion.

1. 1

   You are signing technical decisions monthly. Not one big choice — a stream. A single review answers one; a stream needs someone who remembers the last four.

2. 2

   You have engineers and nobody senior above them. They are not underperforming; they are unrefereed, and every architectural argument escalates to a founder who cannot adjudicate it.

3. 3

   An agency runs your product and you cannot check their work. You are paying invoices you cannot evaluate against a plan you did not write. It compounds monthly.

4. 4

   Something outside is about to inspect you. A funding round, an acquisition, a security review. Someone competent should read your stack before a stranger does.

5. 5

   You keep deferring decisions nobody can make. The tell is a list of “we should look at that” items that has not moved in two quarters. Deferral is a decision, just an unpriced one.

Fewer than two of these, and what you have is a question, not a gap.

Most people who search this question are standing in front of one expensive decision: a quote for a rebuild, an AI proposal they cannot sanity-check, an agency asking for a longer contract. What they need is a few days of senior judgment applied to that one thing, in writing, and then nothing.

The test is not subtle. Write down the decision that made you open this page. If you can state it in one sentence and it has a deadline, buy an answer to it. If you cannot — if it is six decisions wearing a coat — you have the standing gap a retainer exists for.

My own work is arranged that way deliberately. One blocked decision is a Decision Review (2 WEEKS); a whole stack read end to end is a Technical Assessment (3–4 WEEKS). Only the third is a fractional CTO retainer (3-MONTH MINIMUM), because a retainer answers a continuing problem and should be sold as one. What you pay for the smaller is credited against the next within 90 days, and the fee is fixed before anything starts and invoiced 50/50 — how engagements work has the rest. Capacity is ONE NEW ENGAGEMENT PER MONTH; next start SEPTEMBER.

The published ranges do not agree with each other, and pretending otherwise would be the first dishonest thing on this page. Kompella Technologies’ 2026 pricing guide puts US fractional CTOs at $200–$500 an hour and retainers at $8,000–$25,000 a month, while HyperNest Labs’ 2026 cost breakdown reports $150–$200 an hour outside the major US metros. A spread that wide is not a market price. It is a signal that the title covers everything from a weekly advice call to running a department.

The usual comparison is with a full-time hire — the same Kompella guide puts an all-in full-time CTO at roughly $300,000–$500,000 a year including benefits and recruiting fees. That means something only if the full-time hire was ever the alternative, and for most companies asking this question it was not.

Shape matters more than rate. Ask three things of any quote: fixed fee or metered hours; the minimum commitment and the cost of leaving; what you own at the end.

1. 1

   Start with the smallest engagement that answers something. One decision, one deliverable, a fixed fee. You learn how someone thinks by watching them think about your problem, not by reading their case studies.

2. 2

   Put neutrality in the contract, not the pitch. No commission, referral fee or revenue share from any vendor they might recommend. If they will not write it down, you have learned what you needed for free.

3. 3

   Demand written reasoning, not just decisions. The document is the asset, and the only real defence against the handoff problem.

4. 4

   Agree the exit at the start. What has to be true for this to end well? Anyone who cannot describe their own redundancy is selling a permanent dependency.

5. 5

   Buy availability explicitly. Which days, what response time, who covers the rest. “A few days a month” is a hope, not a commitment.