Nobody wants to be the first to say a number, so most agency websites don't. You fill in a form, wait two days, take a call, and eventually hear a figure with no explanation of what drives it — which leaves you unable to tell an expensive quote from a cheap one that has quietly left half the work out.
So: our MVPs start at €5,000 excluding BTW, and most land between there and roughly €25,000. Below is what moves a project along that range, and what to check in any quote you are holding.
Why the range is that wide
The spread is not a negotiating position. It is the honest consequence of "MVP" describing a purpose rather than a size.
An MVP that tests whether people will book a service through a web app is a form, a calendar, a payment, and a way to notify someone. An MVP that tests whether a logistics coordinator will abandon their spreadsheet needs to model the work that spreadsheet does, which is where the real complexity lives — and it usually cannot be discovered from the outside, because it lives in the heads of the four people who do the job.
Both are legitimately MVPs. They are not the same amount of work, and no honest fixed price covers both.
The four things that actually move the number
1. How many user roles there are. This is the single biggest multiplier and the one founders consistently underestimate. One kind of user is one set of screens, one permission model, one onboarding path. Two kinds — a customer and an administrator, say — is not double the work, but it is far more than one and a half times, because every feature now has two answers and the rules for who sees what have to be designed, built and tested.
2. Whether it has to talk to something else. A self-contained product is predictable. The moment it must integrate with a payment provider, an accounting package, a CRM, or worst of all an incumbent system with an API written in 2011, the estimate acquires a tail. Integrations fail in ways that are discovered rather than planned.
3. Whether the data model is known. If you can describe on a whiteboard what things exist in your system and how they relate, you are most of the way to a predictable build. If the answer is "it depends, let me explain the exceptions," expect discovery to take real time — and be glad it did, because the alternative is discovering the exceptions in month three.
4. How finished it has to look. An MVP shown to five design partners can look plain. An MVP that goes on Product Hunt, or in front of an investor, or against a well-funded competitor, needs design work that the first one does not. This is a legitimate reason to spend more; it is also the easiest place to spend more than you meant to.
Notice what is not on that list: the number of features. Feature count is the thing everyone counts and the thing that correlates worst with cost. Ten simple, similar screens are cheaper than three that each require a different way of thinking.
What the entry price actually buys
At the Launchpad end — €5,000 excluding BTW — you get a scoped, functional, deployed product with a single user type, a clear data model, and no complicated integrations. Discovery and planning, core feature development, basic UI, deployment, and two weeks of post-launch support.
That is a real product, not a demo. It is also deliberately narrow, and the narrowness is the point: an MVP exists to answer one question, and every euro spent on something that does not help answer it is a euro spent on an opinion.
How to read a quote you have been given
Whoever wrote it, these four questions separate a considered estimate from a hopeful one.
"What did you assume?" Every estimate rests on assumptions about roles, integrations and data. A quote that does not state them has not made them — it has guessed, and you will meet the guesses as change requests.
"What is explicitly not included?" The exclusions tell you more than the inclusions. Content, data migration, third-party subscriptions, and support after launch are the four most commonly assumed-in by the client and assumed-out by the supplier.
"What happens when we are wrong?" You will change your mind about something in week three. That is the process working. Ask how changes are priced before you need one, not after.
"Who owns the result?" The code, the repository, the infrastructure accounts. If the answer involves a licence, a platform fee, or a hosting arrangement you cannot leave, the sticker price is not the price.
The cheapest MVP is the one you don't build
The most expensive projects we see are not the ones with big budgets. They are the ones that spent four months and €40,000 answering a question that a two-week conversation with eight potential customers would have answered for nothing.
Before you spend anything on building, write down the single assumption that, if false, makes the whole idea pointless. If you can test it without software, do that first. If you cannot, then you know what your MVP is for — and you are now in a position to get an estimate that means something.
Talk it through
If you have a figure in your head and want to know whether it is realistic, tell us what you are trying to build. We will tell you what we would do, what it would cost, and — often enough — that you should build less than you were planning to.