Product
A first product version in 90 days: what belongs in it and what doesn’t
Tomáš DerzsiFounder and Managing Director
- Published
- Updated
- Reading time
- 3 min read
Article contents
The most expensive mistake in custom development isn’t bad code. It’s building features nobody uses. That is exactly why the MVP (minimum viable product) exists: the smallest version of a product that already does real work. It costs a fraction of the budget and tells you whether spending the rest makes sense. The key is deciding what goes in. And above all, what doesn’t.
What a first version really means
It is not a “stripped-down app” or a prototype for show. It is the smallest system that can carry one key process from start to finish for a real user. It runs live, with real data, not in a demo. Otherwise you validate nothing but the design.
What belongs in the first version
One main process from A to Z, for example receiving an order, processing it and shipping it. Plus essential sign-in, basic roles, one critical connection (payments or your main system) and measurement, so you can see whether people actually use it. The goal is clear: the first real user completes the whole task without anyone having to rescue them by e-mail. That applies to an internal system and a web application alike.
What to leave out of the first version
- A second and third role “just in case” and five levels of permissions, when ten people will use the system for the first six months.
- Exports to five formats and making everything configurable.
- Admin screens for things that don’t exist yet.
- Several languages if you sell in one market.
- A mobile app if the website on a phone will do.
- Connections to systems you may never plug in.
Each of these sounds reasonable, and each pushes the launch back by weeks. The rule is simple: if the first user doesn’t need it to complete the main process, it goes into phase two.
Why 90 days
Three months is long enough to build a system that runs live, and short enough that the company doesn’t lose momentum and budget before it sees a result. A first version that takes a year to develop defeats its own purpose: by the time it’s finished, the market has moved and the assumptions are out of date. What doesn’t get validated in ninety days usually won’t be validated in nine months either. You just find out at a higher price.
What 13 weeks look like
- Weeks 1–2: joint design, scope and a clickable prototype.
- Weeks 3–11: development, with a demo of what’s done every Friday.
- Weeks 12–13: launch, measurement and handover.
Regular Friday demos give you around ten points at which you can correct course. The vendor can’t hide behind “we’re still laying the foundations”, and on launch day you don’t get a surprise but a product you watched grow. A vendor who resists this rhythm is telling you more about themselves than about your project.
What happens after 90 days
Launch day isn’t the finish line — it’s where measurement begins. The first month shows what users actually do, and that data, not guesswork in meetings, decides phase two. Sometimes it confirms the direction; sometimes it shows that the most requested feature is completely different from what everyone expected. That is the whole point of a first version: deciding based on behaviour, not on a presentation.
How you know the first version succeeded
Success isn’t measured by it being “done” but by one number agreed in advance: how many real users complete the main process without help in the first month. If, say, 80% complete it and come back, you have a foundation worth building on. If they abandon it after three clicks, you’ve saved most of the budget on a finding that would otherwise have cost you all of it. Both are wins. The only loss is not measuring at all.
The costliest myth: a first version as an excuse for unfinished work
A first version is not permission to ship a broken product. It is a decision to build fewer features, but each one properly. Sign-in must always work, payments must not fail, data must not be lost. These are not “phase-two features”; they are the foundation of trust. When a vendor hides poor quality behind the word MVP, they are not saving you money — just deferring the bill, with interest.
The hardest part of a first version isn’t the code. It’s the discipline to say “not now”. And that is exactly why you need a vendor who can push back, not just nod along.

About the author
Tomáš Derzsi
Founder and Managing Director · Smooth Operator s. r. o.
We work in English and Slovak.
About the company