How I Build 0→1 Products
From a fuzzy hypothesis to a shipped, measurable v1.
Most 0→1 work fails at the seams between research and scoping. The team learns something real, then forgets it the moment a roadmap meeting starts. My job in the first 90 days is to keep the signal alive long enough to ship something the market can react to.
Before a prototype, before a deck, I write the smallest possible problem statement and run it past five people who feel the pain. If three of them rephrase it back unprompted, the problem is real.
Mostly conversations, light synthesis. The risk in 0→1 is over-research — I'd rather get to a prototype off twelve clear interviews than wait for fifty.
Cut the spec until removing one more thing breaks the user's job-to-be-done. That's the MVP. Everything cut goes into a Later list, dated, so the team trusts it'll come back.
PMF signals first, vanity later. For a marketplace, retention of the second-side over week 2 matters more than top-of-funnel; for a SaaS workflow, time-to-first-success beats DAU.
Two-week loops, hypothesis written before the cycle, learning recorded after. Not every cycle ships product — some ship a kill decision.
- How I define PMF signals before shipping (and how I avoid false positives).
- Which metrics make the first dashboard, and which I deliberately leave off.
- How I prioritise the first 10 features against the next 50 — what gets the cut and why.
Scaled a healthcare marketplace from zero to a working two-sided flow. The first version handled fewer features than the spec promised — but it shipped, it took real transactions, and it told us within four weeks where the next investment had to go.
A blank prioritisation board I genuinely use — three columns, every item has a one-line hypothesis and a one-line success signal. Cards graduate from Later → Next → Now only when their hypothesis sharpens.
0→1 is less about prediction and more about reducing how much you have to commit before you learn the next thing. The PM's job is to keep the cost of being wrong cheap.