Budgeting

How much does it cost to build an MVP?

An honest breakdown of what a first product version costs in 2026, what actually drives the number up or down, and how to tell a realistic quote from an optimistic one.

4 min readBy Mayur Kale

Key takeaways

  • Scope drives cost far more than technology, team location or hourly rate. The cheapest way to halve a budget is to cut features, not corners.
  • AI tooling has compressed implementation cost meaningfully, but implementation was never the whole bill, discovery, integration, testing and launch have moved much less.
  • Beware quotes given without questions. A number produced in one meeting is a sales instrument, not an estimate.
  • Budget for the twelve months after launch. A product with no maintenance budget becomes unusable faster than most founders expect.
  • The most expensive thing you can build is the wrong thing well.

Nobody can quote your project from a paragraph, and anyone who offers to is telling you something about how they work. But "it depends" is a useless answer, so here is a genuine attempt at the ranges and, more usefully, at what moves them.

What actually drives the number

In rough order of impact:

Scope. By a distance the largest factor. Every user type, every permission level, every integration and every "and it should also…" is multiplicative rather than additive, because each one interacts with the others. Two user roles is not twice one role; it is roughly three times.

Integrations you do not control. Talking to your own database is cheap. Talking to a twelve-year-old ERP with no documentation and a two-week turnaround on sandbox access is where budgets go to die. This is the single most under-estimated line in most quotes.

Compliance and data sensitivity. Health, financial and children's data carry real engineering cost, audit trails, access controls, retention policies, review. It is money well spent and it needs to be in the budget from the start rather than discovered in month three.

Design ambition. A conventional, clean interface built from a component library is inexpensive. A distinctive, animated, carefully crafted one is a genuine investment. Both are legitimate; they are not the same price.

Quality bar. A product used by ten internal staff and one used by fifty thousand paying customers require different amounts of testing, monitoring and hardening. Deciding this explicitly is much cheaper than discovering it.

Notice what is not near the top of that list: programming language, cloud provider, or whether the team uses AI tooling.

Rough ranges

Ballparks for a competent team building something real, deployed, tested, monitored, with a sign-up flow that works:

ShapeTypical rangeWhat you get
Single-workflow MVP£25k–£60kOne core journey, auth, database, deployed and monitored
Multi-role product£60k–£150kSeveral user types, permissions, admin tooling, reporting
Integration-heavy build£100k+The above, plus systems you do not control
Regulated or high-assurance£150k+The above, plus audit, access control and review

These assume senior engineers and a product you intend to keep. They are not comparable to the lowest quote you can find, and they are not meant to be.

What AI has and has not changed

Implementation is genuinely cheaper than it was two years ago. That is real, and it is reflected in the ranges above being lower than they would have been in 2023.

Be careful about how far you extrapolate, though. The research on AI-assisted delivery is more equivocal than the marketing. METR's randomised trial found experienced developers were 19% slower on tasks in codebases they knew well, while believing they were 20% faster. DORA's 2025 industry data found AI adoption correlating with higher throughput and higher instability at the same time, faster shipping, more rework.

The practical reading for a budget: expect meaningful savings on writing code, expect no savings on deciding what to build, and expect to spend some of the gain back on review and rework. A firm passing on the full theoretical saving is either absorbing the difference or planning to.

Reading a quote

A quote with no questions is a sales instrument. A real estimate follows a conversation about your integrations, your users, your compliance position and what you already have.

Ask what is excluded. This surfaces more than asking what is included. The gaps between quotes are almost always in the exclusions.

Ask for a range with reasons. "£40k–£65k, and here is what determines which end" is a more honest and more useful answer than a single confident number.

Watch for the missing lines. Frequently absent from cheap quotes: testing, deployment setup, monitoring, security review, documentation, handover, and the project management to coordinate it. You will pay for these regardless; the only question is whether they were in the number you agreed.

How to spend less, honestly

  1. Cut to one workflow. Not the best five features, the single one that has to work.
  2. Do things manually at first. Onboarding, support, billing, moderation. Automate once volume forces it, and you will automate the right thing.
  3. Use conventional technology. Novel infrastructure choices cost money twice: once to build, once to hire for.
  4. Defer the admin panel. A database client and a competent person covers more early-stage need than founders expect.
  5. Buy rather than build anything that is not your differentiator. Auth, payments, email, analytics. Every one you build yourself is a permanent maintenance obligation.
  6. Pay for scoping. A week spent getting the specification right is the cheapest week in the project.

The expensive mistake

None of the above matters much next to the main one: building the wrong thing competently.

Every failed product I have looked at was killed by a decision made before any code existed. The budget conversation people should be having is not about rate. It is about how quickly and cheaply they can find out whether anyone wants this.

Frequently asked questions

How much does it cost to build an MVP in 2026?

For a genuinely minimal product, one core workflow, authentication, a database, and a deployed, monitored environment, most quality engagements land somewhere in the low tens of thousands of pounds. Products with several user types, real-time features, payment handling or integrations into legacy systems move into six figures quickly. Any firm quoting before asking about your integrations, your compliance requirements and your existing systems is guessing.

Why do development quotes vary so much for the same brief?

Usually because the firms are quoting different scopes while using the same words. One has assumed a single admin user and a manual onboarding process; another has assumed self-service sign-up, role-based permissions and an audit trail. Ask each to write down what they are excluding, and the quotes usually become comparable, and much closer to each other.

Has AI made software cheaper to build?

Implementation is genuinely cheaper, and that is a real reduction. But implementation was typically less than half the cost of a build. Understanding the problem, integrating with systems you do not control, testing, security review, deployment and handover have all moved much less. Expect meaningful savings on the code-writing portion rather than an order-of-magnitude change to the total.

What should I budget after launch?

Plan for ongoing cost from day one. Hosting and third-party services are the visible part; the larger item is engineering time for the bugs real users find, the dependency and security updates that arrive regardless, and the changes the first month of feedback will demand. A product with no post-launch budget degrades quickly.

Is it cheaper to hire freelancers than an agency?

On hourly rate, usually yes. On total cost, it depends on whether you can supply the technical direction yourself. If you can specify the work, review it competently and own the architecture, freelancers are excellent value. If you cannot, you are buying that judgement from somewhere, and buying it late, after something has gone wrong, is the most expensive option available.

References & further reading

  1. [1]
  2. [2]
  3. [3]

Turn the trend into shipped software

I contract as a senior fullstack and data engineer, building the production systems your customers and your reporting depend on.

Explore services↗