Flowdesk

Demand was already proven: twenty-three service companies had confirmed in writing that they would pay. The product did not exist — not a line of it. Investor money closed in ten weeks, and what had to be shown was a working service, not slides.

The danger was not the deadline. In eight weeks it is easy to bolt together something you throw away and rewrite six months later — and then the fast start costs double.

The first week went into cutting, not building. Of twenty-one planned features, six made the first version — the ones without which nobody would pay. The rest were not thrown away: they went into a plan with dates, so nothing was lost and the founders could see when it would arrive.

Separation between companies and staff roles — who sees what, who can change what — went in from the start. That cost four extra days up front. It saved months when the first twenty-person client arrived: nothing had to be redone.

Getting to know the service became part of the service. A new user goes through four short steps and lands in a workspace that is ready to use rather than empty. That removed the most common reason people never come back after signing up.

Every feature was judged by one question: would anyone refuse to pay without it. If not, it went into the plan for later.

A mobile app, customisable reports and connections to two popular CRMs did not make the first version. None of the twenty-three prospects named them as a condition of buying — and together they would have added at least a month.