BenjaminSchleier
Blog

Start-ups

How small should your MVP be?

Benjamin Schleier · November 5, 2026 · 4 min read

With a new software idea, the list of possible features usually gets written quickly. The harder question is what you actually need for a first test. I have developed products of my own and worked with start-ups, and keeping the scope of the first version deliberately small has served me well.

1

An MVP answers exactly one question

Many feature lists describe the finished product more than the first test. An MVP describes an experiment. So start by writing down the assumption your business depends on, phrased as a sentence that could turn out to be false. “Tax advisors will pay to get their clients' receipts pre-sorted” is that kind of sentence. “The market for tax software is growing” is not, because no experiment can ever refute it, so it helps you with nothing. Everything that tests the first kind of sentence goes into the first release, and everything else waits. That cut hurts, because the founders' favorite features rarely make it.

2

Sell first, then build

A first test does not have to start with programming. Talk to people who supposedly have the problem and ask about the last time it came up. What did they do about it, and what did that cost them? Write the answers down word for word. Someone who describes a case from last week and has already cobbled together a spreadsheet as a stopgap is a potential customer. “Sounds interesting” is pure politeness. At the end of the conversation, ask for something that takes effort, like a pre-order or a fixed date for a test. Praise is free. That is exactly why it tells you nothing.

The next step is a single page that describes your offer and has one button. Run a few ads on the search terms your customers would type, and count who signs up. You learn even more if you deliver the service by hand. Your first customers get their result by email, while behind the curtain you sit with a spreadsheet. Nobody needs to know how manual it is yet. That does not scale to a hundred customers, and it does not need to. After a handful of jobs like this, you know which steps software has to take over and which ones nobody would miss.

3

What belongs in the first version

Once it is settled that something gets built, write a short list of the features you cannot do without. One type of user and one task is all a first version needs. Roles and permissions go, as long as there is only one kind of user, and so does the settings page. You write invoices by hand for the first few months. A payment integration pays off from the moment that gets tedious. At first, the admin area is a direct look into the database. The phone app waits until people are actually coming back in the browser often enough to want one. Every one of these cuts feels unprofessional. Your first users will not notice any of them, because they came for the one function that solves their problem.

4

Where you must not cut corners

Small and sloppy are two different things. The one function the whole thing is about has to be good. Otherwise you are testing your implementation and learn nothing about the idea. Handle your users' data with care from the first day, which includes an encrypted connection and a backup you have actually restored once. Build in a simple measurement that shows you who signs up and who comes back after a week. And keep a short line to your users. An email address that reaches you personally is enough. Read every reply yourself, because in this phase every complaint is worth more than a compliment, and a bug report from a real user is a gift.

5

Count the build time in weeks

Give the first version a deadline in weeks, and write it down. If the date slips, you cut scope. The question then becomes what really has to be in the first test and what can wait. How many weeks are realistic depends on the product. A booking tool is faster to build than something that handles payments or processes health data. If the uncertainty is high, build a prototype first that tests only the riskiest assumption. It costs a fraction of the full build and tells you whether the rest is worth it.

6

Define success before launch

After launch, numbers are easy to read in different ways. So write down beforehand which result means you carry on and which one means you turn back. That might be a number of paying customers after the first month, or the share of testers who return in week two. How high the bar sits depends on your market. What matters is that you set it before the first data comes in. Put it in writing where the whole team can see it. If the result stays below it, that is an answer too. You change the target group or the offer and test again, with the money you did not sink into six months of development.

  • MVP
  • Start-ups
  • Product development
  • Validation

Let's talk.

Write me two sentences about what you need. Or pick a time for a 20-minute video call.

What is it about?

I use your details only to reply to you. More in the privacy policy.