MVP development cost in the UK: price bands for 2026, and what moves them

Building an MVP in the UK costs roughly £15,000 to £80,000, in 4 to 12 weeks, depending on how much you build and who builds it. Kyln's product and platform builds start at about £20,000, priced after a £1,500 review rather than guessed from a feature list. This guide sets out the 2026 price bands, what each one buys, who charges what, and where the number actually comes from. One declared interest up front: Kyln builds MVPs. The market's numbers are here next to ours, so you can judge both.

The short answer

The bands are wide because "MVP" covers everything from a weekend demo to a fundable platform, and because the same build costs very different amounts depending on who makes it and how much gets thrown away later. Both are unpacked below.

What you are actually paying for

An MVP is not the screens. The screens are the cheap part, and they are cheaper than ever now that AI tools draft them in an afternoon. What costs money is everything behind them: a data model that will not need ripping out, authentication and permissions that hold, payments wired to something real, deployment that can roll back, and enough testing that a change does not quietly break the last one.

That is the last 20% of the work, and it is where most of the price and almost all of the risk live. A build that skips it looks finished and behaves like a prototype the first time a stranger uses it. Paying for an MVP is paying for the parts a demo lets you ignore.

What UK builders charge

Checked July 2026, advertised ranges. Prices are "from" prices and hide different exclusions, so compare what is actually included, not the headline.

The spread on the same build is real: a product that is £35k offshore can be £55k with a hybrid team and £85k at a UK agency. The cheaper number is not automatically the worse deal, and it is not automatically the better one. What decides the true cost is how much of the work has to be redone before the thing can carry real users, and that does not show up on the quote.

Why the range is so wide

Four things move an MVP's price more than anything else.

• Scope. One workflow done properly is a different job from five done partially. The single most effective way to lower the cost is to build less, first.
• Seniority. A senior engineer costs more per day and less per outcome, because the expensive mistakes in an MVP are architectural, and those get made early or not at all.
• Who carries the risk. A fixed price moves the risk to the builder and buys you certainty, but only holds if the scope is tight. Day rates are more honest while the requirements are still moving.
• How much gets thrown away. The cheapest-looking build is dear if it has to be re-architected before launch. The real figure is build cost plus rework, and rework is invisible at quote time.

The AI-era twist nobody prices in

Most MVP cost guides still price by the screen, and that model is breaking. AI coding tools now draft something like 80% of a feature on their own, so the interface is no longer the expensive part. The cost has moved to the last 20%: the judgement to make the generated code safe, and the architecture, security and testing it skips.

Two things follow. First, a cheap AI-built prototype is not a cheap MVP. Tools like Lovable, Bolt and v0 get you to a working-looking demo for very little, then the same auth, data and deployment work is still waiting, and an unreviewed build can cost more to make safe than a clean start. Second, the right question is no longer "how many screens" but "how much judgement to make this real". A build partner that leads with AI speed and no plan for the last 20% is quoting for a demo, not a product. If you already have a prototype, the free readiness scorecard is a two-minute read on how far it actually is from something you can sell.

Kyln's numbers, in full

Our own figures, so the market table has something concrete to stand against. Every price is public, and the same list is on the pricing page.

The order matters. We price the build after the review, not from a feature list emailed in cold, because a number guessed before anyone has read the problem is a number that gets renegotiated. The review costs £1,500 and buys the honest scope; the build cost follows from it. Where a founder needs ongoing technical direction rather than only delivery, that can run as fractional-CTO direction from £3,000 a month. Ownership stays with you throughout: the code, the product and the accounts are yours.

Should you spend that at all?

Honestly, often less than you think, and later than you think. The most common expensive mistake is building the whole idea when the first version only needed one workflow to prove the point. A tighter first build is cheaper, faster and easier to change once real users tell you what was wrong, which they always do.

The founders who get the most from an MVP budget are the ones who scope it down before they spend it: one buyer, one workflow, one number to move. If you have committed budget and no senior technical lead to make those cuts, that is exactly what the review is for. Spend the £1,500 to find out what not to build, then spend the rest on building it once.

Common questions

How much does it cost to build an MVP in the UK?

Roughly £15,000 to £80,000 for a real MVP, with lean single-flow builds landing around £15,000 to £30,000 and standard multi-flow products £25,000 to £60,000. Clickable prototypes are cheaper (£5,000 to £15,000); funded v1 platforms run higher. Kyln builds start at about £20,000, priced after a £1,500 review rather than from a guessed feature list.

How long does it take to build an MVP?

A credible first version is 4 to 6 weeks if the scope stays narrow: one core workflow, done properly. Add custom reporting, roles, SSO or several integrations and it is 8 to 12. The timeline is set by the number of real flows and the integrations, not by the number of screens.

Is it cheaper to build an MVP with AI tools like Lovable or Bolt?

Cheaper to a demo, yes. Cheaper to a product, often no. AI tools get you a working-looking prototype fast, but the auth, data model, security and deployment underneath usually still need building, and an unreviewed AI build can cost more to make safe than a clean start. Run the free scorecard, or start with a review, before treating a prototype as most of the way there.

Should I use an offshore team to save money?

The sticker price is lower and sometimes that is the right call. The cost that moves is the management and the rework: an MVP that has to be re-architected before it can take real users is not cheap, whatever the day rate was. Judge the total, including your own time, not the headline.

What is the difference between a prototype and an MVP?

A prototype proves an idea is worth building; an MVP is the smallest version real customers can actually use and pay for. The gap between them is auth, data, security, deployment and the dull operational work, which is also most of the cost. Confusing the two is the most expensive mistake in this whole market.

Can I get a fixed price for an MVP?

Yes, when the scope is tight. A fixed price gives you certainty, but only if the requirements have stopped moving; if they are still shifting, day-rate delivery is more honest than a number that will need renegotiating. Kyln prices the build after a review, so the figure is based on the real scope rather than a guess.

Kyln's build and care pricing sits together on the pricing page, the founder route is on the for founders page, and if what you have is a prototype rather than a plan, start at the rescue hub or run the readiness scorecard.