What a fractional CTO actually costs in the UK

Enough people in this category now publish a price that “it depends” has stopped being an acceptable answer. Here is the range, where it comes from, and where I sit in it.

Search for what a fractional CTO costs in the UK and you’ll find page after page explaining what a fractional CTO is, gesturing at “it depends”, and inviting you to book a call. That’s a sales funnel. So: the answer first, then the part that matters more, which is what the number buys.

The going rate

Three models are common in the UK. They price differently.

Day rate. £800 to £1,600 a day. At the lower end you’re buying a senior engineer who also advises. At the upper end you’re buying someone who has run a technology function and will make decisions you can’t make yourself. This band is the one I see quoted and paid in the UK market, not the output of a survey, so treat it as a working benchmark and check it against whoever you talk to. Day rates are the least published of the three models, which is worth noticing on its own.

Monthly retainer. £3,000 to £8,000 a month for SME and startup engagements, with a cluster around £4,000 to £6,000. That band is what I see quoted and paid. Treat it the same way as the day rate above: a working benchmark to check against whoever you talk to, and nothing like a measurement of the whole market. Some providers do put this on the page. fcto.uk lists an advisory tier at £3,000 to £5,000 a month and an active CPTO tier at £8,000 to £12,000, excluding VAT. Hurricane Studio publishes $5,000 to $30,000 a month and says most engagements land between $10,000 and $20,000. Both checked August 2026. This buys a cadence and not a count of days: a standing commitment to own the technical direction, sit in the meetings that need a technical decision in the room, and be reachable when something breaks.

Productised or subscription. A smaller group sell “unlimited technical input” for a flat monthly fee, sometimes with equity or revenue share instead of cash. This is a different product from the other two. You’re buying availability instead of time, which works well when the need is unpredictable and badly when it’s heavy.

For comparison, hiring the role outright. Robert Half’s 2026 UK salary data puts a CTO at £120,000 at the 25th percentile, £160,500 at the median and £201,500 at the 75th, checked August 2026. The spread matters more than the midpoint: a £120,000 CTO and a £200,000 CTO aren’t the same hire. On top of any of those sit equity, employer’s national insurance, pension, recruitment fees, and the three to six months it takes to fill the seat.

What I charge

I have just spent a section telling you what other people cost, so here are mine.

The diagnostic is £7,500. Two weeks, fixed, and the price doesn’t move with the size of your business. A complicated business makes the narrowing harder, and the narrowing is what you’re paying for. The fortnight stays a fortnight. It’s the one price on this page that isn’t a rate: it’s priced on what it settles, and two weeks is the window, not a count of days.

Day rate is £1,000. One number, not a band, because a band turns every conversation into a negotiation about which end you’re at. It sits in the lower half of the market band above, and that’s deliberate: the decision itself, owned with its consequences, is what the diagnostic and the retainer sell, and both are priced as a defined thing at a fixed number rather than as a premium day rate for an undefined one.

Retainers start at £2,000 a month for a support retainer of two days a month, for a system that’s live and needs looking after. Beyond that, £4,000 a month buys one day a week and £7,500 a month buys two. The support retainer comes out at £1,000 a day. There are 4.33 working weeks in a month, so the weekly ones come out at £924 and £866 a day. All three sit inside the band above, and the rate falls as the commitment rises, which is the shape you should expect. Ask anyone quoting you a retainer to show you the same sum. Be careful if they will not.

The weekly retainers cost less a day than my £1,000 day rate because you’re committing, and the support retainer costs the same because two days a month is too little to discount. Beyond the days, a retainer reserves a slot I can’t sell to anyone else, and it buys you being able to reach me between the scheduled days: email or chat, for a decision that can’t wait. The aim is a considered reply within one working day on a weekly retainer, or two working days on the support retainer, because two days a month can’t hold the same slot. Those are targets I work to, not a guaranteed service level, and as-needed work carries none.

No minimum term. Retainers run month to month, with 30 days’ notice either way. I’d rather you stayed because the arrangement is working than because you signed something in March. A retainer that only survives on a lock-in had already stopped being worth paying for, and both of us would know it.

More than two days a week is possible but I don’t publish a price for it, because three days is most of one person’s week and I’d rather say that than pretend otherwise.

If you need less than two days a month, I work at my £1,000 day rate as and when: booked in half days and invoiced for what is used. Nothing is reserved, so there’s no reply promise.

Build work is quoted, never rated. I’ll quote a build the moment there’s something to quote against, and a brief I can quote against has four things in it: what done looks like, including how fast, how available and how secure it has to be, and what is deliberately out of scope; representative data with its shape, its volumes and the awkward cases, or the interface contracts that carry the same detail; where the code lives now, what it integrates with, how it gets deployed, what has to be migrated, and any regulatory constraints; one named person who can settle an open question inside an agreed time. Hand me those and the quote is free and you don’t need the diagnostic at all. Hand me three of them and what you get is a range with the assumptions written next to it, which is an honest answer and not a quote. Closing those assumptions is what the diagnostic does. What I won’t do is put a number on a build I can’t see the shape of: a price given without a specification is a guess wearing the costume of a quote, and you’d be right not to trust one.

All prices exclude VAT. I’m VAT registered, so VAT is added to every invoice. Whether you can reclaim it depends on your own business, and your accountant will know.

If those numbers are wrong for you, the useful thing is that you found out from a page instead of from three emails and a call.

What the range is really measuring

The spread between £800 and £1,600 isn’t a quality gradient. It measures how much of the decision you’re handing over.

At the bottom you’re buying execution with judgement attached: someone who will build the thing well and tell you if the plan is wrong. At the top you’re buying the decision itself, from someone who will own being wrong about it. Those are different products, and the mistake I see most is paying for the second while scoping the first.

The signal to watch for is what happens when you ask a question they don’t already know the answer to. Execution says “what would you like me to do?”. Ownership says “here is what I think, here is what would change my mind, and here is what I need from you to decide”.

What it isn’t buying

Three things. Being clear about them now saves the awkward conversation later.

It isn’t buying throughput. A fractional CTO two days a week doesn’t produce two days a week of code plus the leadership. If you need volume you need engineers, and a fractional CTO is the person who tells you how many and what shape. Not the person who replaces them.

It isn’t buying availability in a crisis, unless you’ve paid for that explicitly. Someone with four clients has four crises, and the honest ones tell you what their response commitment is instead of implying it’s total.

It isn’t buying institutional memory. This is the real cost of the model and almost nobody prices it in. A full-time CTO accumulates context about your business every day for years. A fractional one accumulates it in the hours you pay for. You can manage that gap, but only by writing things down, and most companies do not.

The cheapest expensive mistake

The most expensive thing in this category is building the wrong thing competently. The day rate is small change beside it.

I watched a business spend serious money on a system that worked exactly as specified and solved a problem it didn’t have. The specification had been written before anyone established what the real constraint was. Nobody did anything wrong. Everyone executed. The money is still gone.

That’s why I sell a fixed-price diagnostic instead of starting with a build. Two weeks, £7,500. I triage what is costing you, establish why from your own systems, take the riskiest assumption in the plan and test it, and then we decide what to build and what it costs.

That third step is the one that matters and the one most people skip. It takes one of two shapes, and we agree which before you commit. Where the risk is that the thing can’t be built, or can’t be built the way everyone assumes, you get software that runs: the assumption it tests, the acceptance test it has to pass, the result, what it deliberately doesn’t do, and the source. Where the risk is in the diagnosis instead, you get a named experiment against your own data with the threshold agreed before it ran, and the finding it settles, including the finding that stops a build. What I won’t do is promise software and hand over a longer document. There is a worked illustration of the whole deliverable, written out for a business that doesn’t exist, if you want to see the depth before you decide.

It’s priced as its own piece of work instead of as a deposit on a build, because it’s worth what it’s worth whether or not I build the result.

It’s fixed and not daily because of what a day rate would do to my incentives. Paid by the day to decide what to build, I get paid more for taking longer to decide. Paid a fixed fee, I want the right answer and the exit.

And if the answer is that you shouldn’t build anything, that’s the answer, and it’s worth a great deal more than the fee.

How to work out what you should pay

Stop benchmarking the rate. Work backwards from what a wrong decision costs you.

If the decision is worth £30,000, say a small internal tool, a workflow automation, a choice between two vendors, then a few days of senior input is proportionate and a long retainer is not.

If the decision is worth £300,000, say a platform rebuild, an architecture you’ll live with for five years, whether to build or buy the thing your business runs on, then the difference between good advice and adequate advice is worth many multiples of the fee. Shopping on rate is a false economy.

If the decision is worth your company, the technical direction of a product you’re raising against, then you aren’t buying advice at all. You’re buying someone who will own the outcome with you, and that conversation is about alignment and commitment, not a rate card.

The question worth asking instead

When you talk to anyone in this category, ask them what they would tell you not to do.

Everyone can describe what they would build. The people worth their rate can tell you which parts of your plan they think are wrong, and what evidence would change their mind. It costs nothing to ask and it tells you almost everything.

If they can’t do it in the first conversation, the rate isn’t your problem.