How I actually work.

Phase 1: Understand

The first 2–3 weeks of any engagement are mostly reading and listening. Existing artefacts (Figma files, docs, GitHub PRs, Slack archives, customer transcripts). Conversations with the people who care about the problem most. I take a lot of notes, almost all of which I throw away.

I’m looking for two things: the explicit problem (what people say is broken) and the implicit problem (what’s broken that nobody is naming). They’re almost never the same. The second is usually more interesting.

Phase 2: Frame

I write a memo. Always. 1,500–3,000 words. The memo describes the problem as I now understand it, the constraints I think I’ve identified, three or four candidate directions, and which one I’d advocate for. The memo is the artefact — not a Figma file. The Figma file comes later.

I share the memo with the smallest possible group of people who can give me useful feedback. I revise it. Then I share it more widely. The revision cycle is where the real work happens.

Phase 3: Make

Now Figma. Now sketches. Now prototypes. The making goes fast because the thinking is mostly done. If I’m struggling at this phase it’s almost always because I cut Phase 2 short.

I push to a shared workspace daily. I want everyone watching me work to be able to see the trail of my thinking. The bad ideas stay visible alongside the good ones. That’s the point.

Phase 4: Pressure-test

Before I commit, I look for the strongest reasonable argument against the chosen direction and write it down as if I believed it. Sometimes I find it convincing and change course. Sometimes I don’t and I have a much better answer when someone else raises it later.

Phase 5: Hand off + follow

I don’t love the word “handoff” — it implies the work is over. I think of this phase as ongoing pairing with whoever is going to ship the work, for as long as they want me. Most engagements include a 4–6 week follow-on after launch where I’m available for the inevitable questions.