Fractional CTO: what it is, what it costs, and when to hire one

A fractional CTO is a part-time chief technology officer. The role owns your technology decisions for one to three days a week: the architecture, the engineering team, what gets built and in what order, and which vendors you depend on. You get the judgment of someone who has done the job before, without the salary, the equity or a hiring search. I have been a co-founder and CTO seven times, and I now also work as a fractional Chief AI Officer. That second part is why this page exists. A growing number of companies go looking for a fractional CTO when the question they have is about AI. The two roles are not the same job.

What does a fractional CTO do?

The same things a full-time CTO does, in fewer hours, with more of the time spent on decisions and less on meetings. In practice that means owning five things.

• The decisions that are expensive to reverse. Where the data lives, how the systems connect to each other, what runs where. Most technology choices are cheap to change. A handful are not, and the job is knowing which is which.
• The team. Who to hire next, how work moves from an idea to production, and being the person whose "no" is final.
• Build or buy. Which vendors you depend on, and what it would cost to leave each of them.
• The roadmap in engineering terms. What is realistic this quarter, what is not, and what has to happen first.
• People outside the company. An investor's technical due diligence, a large customer's security questionnaire, a board asking whether the platform can handle growth.

The time commitment is usually one to three days a week. The first month takes more days, because that is when the decisions get written down.

One example of what the role is for. On a retail platform I led for nineteen months, with a team of seven, the decision that mattered most was not a technology choice. It was keeping the path that takes a sale separate from the path that reports on sales, so that reporting could never slow down checkout. Nobody would have noticed it missing until the first busy season. By then it would have cost a rebuild. That is CTO work: making a small decision early, so you do not pay for a large one later.

Fractional, interim or full-time CTO?

Three arrangements, and they solve different problems. Part-time CTO and virtual CTO are other names for the fractional one.

The label matters less than one question: does the person have the authority to say no? A fractional CTO who cannot stop a bad decision is only giving advice.

Do you need a CTO or a Chief AI Officer?

Most pages comparing these two roles are written by someone selling one of them. I do both jobs, so I have no reason to steer you toward either. Here is how I tell them apart, by the problems the company has now.

The case nobody writes about is a company that already has a CTO, and still has an open question about AI. Can your existing CTO own AI? Three tests answer that.

1. Have they shipped a system where a model's output reached a customer? And did they measure how often it was wrong? If not, they will be learning at your expense.
2. Do they have the hours? A CTO running a product and a team rarely has a spare day a week. AI decisions need one.
3. Will they say no to a build their own team wants? The person deciding what to leave alone should not be the person whose job is building.

If all three are yes, give your CTO the budget and the authority. You do not need me. If any is no, the usual answer is a fractional Chief AI Officer working alongside the CTO, not above them.

When a company has both, the split that works is simple. The CTO owns how things are built and run. The Chief AI Officer owns whether a use of AI is worth doing, what data it may touch, and how anyone will know it is working. Where the two overlap, the CTO has the final word on the platform and the Chief AI Officer on the decision record. Write that split down on the first day, or it gets argued about in the third month.

When you need neither

Four situations where hiring either one is the wrong move.

• You need engineering hours, not judgment. If the plan is clear and the team is short, hire engineers or a contractor. An executive on a retainer will not do the engineering work in your backlog.
• You have not decided what you are building. No CTO can design a product the owner has not defined. A fractional one will bill you while you decide.
• The question is small. One vendor decision or one architecture review needs a short fixed engagement with a written answer, not a monthly retainer.
• You need someone fifty hours a week. That is a full-time hire. Trying to fill it part time gets you an overworked contractor, and nobody available when they are busy elsewhere.

What it costs to hire the wrong one

Nobody publishes this, so here is the arithmetic. The most solid published market rate for a fractional CTO is $10,500 to $15,000 a month at fifteen hours a week. That comes from a placement marketplace's own data. Three months with the wrong hire at that rate is $31,500 to $45,000 of retainer. The retainer is the smaller cost. The larger cost is the three months lost, which happens in one of three ways.

• A generalist CTO hired for the AI work. They pick a model provider and a framework and ship a demo. Nobody builds the test set that tells you when the model is wrong. You find out months later, from a customer.
• An AI strategist hired as the CTO. The decision record is excellent and nothing ships, because nobody owns the platform the decisions are meant to run on.
• Either one, when you needed engineers. Three months of useful meetings, and the backlog is unchanged.

The fix costs almost nothing: decide which role you need before anyone starts a retainer. That is most of what a first conversation with me is for.

What are the risks of hiring a fractional CTO?

Real ones, and each has a mitigation you can insist on before signing.

• A slow start. Someone who works with you one day a week learns your systems more slowly. Book more days in the first month, and write every decision down.
• No authority. Write down who decides what before any work starts. On my engagements that split is a table anyone can read, and it applies for the whole engagement.
• Advice without delivery. Insist on something real every week, that a person outside the team can open and use. Progress nobody can inspect does not count.
• Not there when it breaks. One to three days a week is not on call. Agree on day one who responds when something breaks, and how. If the business needs someone on call at all hours, that is a full-time role.
• Overstaying. A company past roughly fifteen to twenty engineers usually needs a full-time CTO. A good fractional one plans the handover to that hire rather than resisting it.
• Split attention. Every extra client means less of that person's time for you. Ask how many engagements they take on at once. I take one or two at a time.

What does a fractional CTO cost?

Published figures vary widely, and only two rest on a disclosed sample. Both come from placement marketplaces, which earn from the placements, so read them as the market's asking price. One puts the middle half of hourly rates at $175 to $250, and $10,500 to $15,000 a month at fifteen hours a week, from 787 profiles over ninety days. The other quotes a day rate of $1,649, from 68 verified rates. Google's own summary of the search quotes $150 to $500 an hour, $3,000 to $15,000 or more a month, and $235,000 to $425,000 or more a year, plus equity, for a full-time CTO.

What changes the number is how much of the week you need, whether the role includes building, and how many teams it covers. I price both roles the same way: a monthly retainer for an agreed number of days, quoted as a fixed figure after the first conversation.

How I work, in either role

The same way, whichever role it is. The owner's decisions are written down before any code, under rules for what may be touched in live systems. The split of authority is written next to them. Something real ships every week. It ends with a handover written as a contract, so the next decision does not need me.

My last engagement ran eight months. It produced about 2,900 commits against 111 tracked issues, with four automatic checks on every change. The product passed 127 unit tests and 179 endpoint tests before a single vendor credential existed. The handover ran to ten documents across four codebases. That is what the role looks like when the person in it also builds.

The rest of the record is on the bio page: twenty years in enterprise architecture, seven companies built and sold, and 8,202 contributions in the twelve months to 14 September 2026, across retail, trading, security, video, robotics and inference work, including a 26-billion-parameter model measured on a desktop with no graphics card.

Common questions