PHASE ONE — DISCOVERY.

//What happens#

I read the existing code if there is any, talk to whoever is closest to the users, and look at whatever analytics exist. Then I build one thing — the hardest interaction, or the screen with the most unanswered questions — as a rough prototype. Not to keep, but because a prototype produces information that no amount of discussion does.

I also go looking for the constraints nobody mentioned. The CMS that can't express the content model the design assumes. The legacy endpoint that takes eight hundred milliseconds and can't be changed this year. The compliance requirement that rules out the obvious approach. These exist on every project and they're never in the brief, because the people writing the brief have stopped noticing them.

//What comes out#

A written scope with phases and an estimate per phase. An explicit list of assumptions, each one something that would change the estimate if it turned out to be false — so when one does, we can both see exactly which number moves rather than renegotiating from scratch. And a list of things I recommend cutting, which is usually the most contentious part of the document.

//Why it's paid#

Because it's the phase that determines whether the rest of the money is well spent, and because free discovery quietly incentivises optimism. Twice now discovery has ended with me recommending the client not proceed. Both were the right call and both were only possible because I wasn't hoping to win the build.

What comes out of this phase feeds straight into the build, and the constraints agreed here are what each of the case studies is organised around.