The simplest version that generates learning
Identify the 1-3 core features of your product. Build only those, even if done poorly. Goal: show the idea to customers, not impress them.
Actionable metrics vs vanity metrics
Ignore views and downloads. Measure: % customers who repeat purchase, customer acquisition cost, retention rate. One metric = one specific behavior.
Get out of the building and talk to people
Interview 10 potential customers this week. Ask: "What's your biggest problem with [area]?" Don't sell, just listen and take notes.
Iteration speed beats perfection
Establish 1-2 week cycles: build something, measure response, learn what works. Reduce cycle time each time.
Failure is learning
Every month ask: "Am I building what customers really want?" If the answer is no for 3 months, plan a radical pivot.
Innovation accounting
First week: establish starting numbers (e.g., 2% conversion). Second week: test small changes. Don't optimize until you have baseline data.
Test removable hypotheses
For each new feature, write: "I hypothesize that [feature X] will increase [metric Y] by [Z%] because [reason]." Test for 2 weeks, then decide.
Group users by acquisition period
Divide customers by signup month. Compare behaviors: does the January cohort behave better than March? Why?
Sticky, viral or paid
Decide: do you grow by retaining customers (sticky), word of mouth (viral) or buying traffic (paid)? Focus everything on ONE engine at a time.
Validated learning > task completion
At week's end don't ask "What did I finish?" but "What did I learn about my customers that I didn't know Monday?" Document every insight.