The six-week rule, and what it cost us to learn it

In 2018 we took on a project we estimated at nine weeks. It shipped eleven months later. Nobody behaved badly and nothing went wrong in a way you could point at. It just kept not being finished.

We did the usual post-mortem and reached the usual conclusion: we had underestimated. So we started padding. The next quote was fourteen weeks for work we thought was nine, and that one took seven months.

The estimate was never the problem. The problem was that an open-ended timeline is an invitation, and every project accepts it. Someone always has one more idea, and when there is time left on the clock there is no reason to say no.

What we changed

We stopped estimating and started constraining. Every engagement is now quoted as a fixed six-week block at a fixed price. If the work does not fit in six weeks, the work is wrong — not the number.

That sounds like a scheduling change. It is actually a scoping change, because it moves the hard conversation to week one, where it is cheap, instead of week nine, where it is expensive and personal.

The deadline is not there to make us work faster. It is there to make us decide sooner.Something Sam said in 2019 that we have been repeating since

What it looks like in practice

The first two days are workshops with everyone who can say yes. We leave with two documents: a one-page scope, and a list of everything we have agreed not to build. The second list is usually longer and it is the one people re-read.

• If a new idea arrives in week three, it goes on the not-doing list. It can be the first thing in the next block.
• If we are behind in week four, we cut scope, never quality. The scope was always the variable.
• If we finish early, we ship early. We have never once been asked for the difference back.

The part we did not expect

Clients like it more than we thought they would. A fixed price and a fixed date turns out to be worth more to most founders than the theoretical possibility of getting a slightly bigger v1 for the same money.

Our average project length dropped by about a third. The products got smaller. As far as we can tell from what happened after launch, they also got better.