How to scrappily hire for, measure, and unlock growth

Crystal Widjaja, chief product officer at Kumu and former growth leader at Gojek in Southeast Asia, shares how she built growth at scale in the region. The conversation covers scrappy growth experiments, retention benchmarks, growth team structure and hiring, instrumentation and analytics, and her nonprofit Generation Girl.

01Key learnings

• Run Wizard of Oz style experiments before building: use WhatsApp groups, screenshots, or manual processes to validate demand before engineering investment.
• Experiments are worth running even with around 30 users; the trends stay similar, and precision improves with more data, while zero data teaches nothing.
• Define retention concretely as cohort retention; for a free weekly product, expect roughly 60% of the prior week's users to return and cohorts to flatten within two to three weeks.
• Find the step just before conversion that causes drop-off, such as trust or a painful form, and fix that constraint rather than chasing a flashy new feature.
• Layer sure wins, like speeding up a slow search that drops conversion, alongside speculative bets on new features.
• Map growth using the physics of your market, product, model, and channels first, then change one variable at a time to find levers like using drivers as salespeople.
• Treat metrics as insights only when they explain why behavior happens and lead to action; instrument events with rich context properties, not just bare event names.
• Hire a first growth person who is strong in statistics and experiment design, and bias toward quick, hacky tests over long tool integrations that burn months.

02Frameworks mentioned

Summary and takeaways written by PM Atlas; quotes are short excerpts. © the original author.