Scoping a first version that survives contact with users

Almost every founder we meet arrives with a feature list. It is usually thoughtful, usually correct about what the product will one day need, and usually a year of work. Our first job together is to find the version that takes a quarter.

That is not about building less for its own sake. It is about getting to the moment where real people use the product, because that moment answers questions no amount of planning can.

Name the one question

A good first version exists to answer a single question. Will treasurers move their savings group records off paper? Will clinics pay monthly for appointment reminders? Will hauliers accept loads from a stranger on an app?

When the question is clear, cutting becomes easier. Every feature is judged by one test: does the first version need it to answer the question? Reports, admin panels, referral schemes and dark mode rarely pass.

Cut breadth, not depth

The instinct under time pressure is to build every feature a little. We prefer to build fewer features completely. A savings app that handles contributions perfectly — including the awkward cases like partial payments and late penalties — teaches you something. One that does contributions, loans, chat and investments shallowly teaches you nothing, because every failure has five possible causes.

Keep the manual parts manual

Some things that will one day be automated can start as a person with a spreadsheet. Onboarding calls instead of a self-serve flow. A weekly export instead of a dashboard. An operations team member matching loads by hand while the algorithm is being designed.

Manual work is slow, but it is also a research method. The person doing it learns exactly what the automation will need to handle.

Decide what you will measure before launch

Finally, write down what success and failure look like before the first user arrives. Three numbers are enough. Without them, every launch looks like a qualified success, and the next quarter's roadmap is decided by whoever argues best.

A first version is not a small version of the final product. It is an experiment with a user interface. Scope it like one.