What actually drives the cost of an MVP

Why nobody quotes a number up front

Because the same sentence — "an app where customers place orders" — covers a two-week build and a six-month one, and which it is depends on things the sentence does not say. A quote given before those are settled is a guess dressed up as a price.

The four things

Who has to agree. One founder who decides is fast. Three stakeholders with different ideas of what the product is will cost more in revisions than any feature will in code, and the revisions arrive late.

What it has to connect to. A self-contained app is predictable. The moment it has to talk to an accounting system, a payment gateway or a warehouse that was built in 2011 and has no documented API, the estimate widens — not because the integration is hard, but because nobody knows what is in there until someone tries.

Who logs in. One kind of user is one set of screens. Add an admin, then a partner, then a reviewer, and you have multiplied the screens, the permission rules, and everything that can go wrong between them.

Whether it handles money or personal data. Both bring requirements that are not features: audit trails, who may see what, what happens when a record is wrong. These rarely appear on a feature list and they are not optional.

What does not move it much

The number of screens, the choice of framework, whether it is a web app or a mobile app. These are the things people ask about first and they are mostly noise against the four above.

Getting to a real number

Scope the thing before pricing it. We do a short paid discovery, come back with a scope and a timeline, and only then a figure — and the figure holds because the unknowns were taken out first. Tell us the problem and we will say what we would need to know.