At Duolingo, the Growth team has a backlog of 18 ideas to improve free-to-paid conversion before the back-to-school season, including paywall copy changes, onboarding prompts, referral incentives, and pricing-page redesigns. You are the PM leading execution for a 10-person cross-functional team: 4 engineers, 2 data scientists, 1 designer, 1 analyst, 1 growth marketer, and 1 QA lead. Leadership wants a decision framework within 3 weeks so the team can stop debating and start shipping.
The core question is which ideas deserve a full experiment versus a lightweight analysis such as historical cohort review, funnel analysis, or a quick prototype readout. The decision matters because the experimentation platform is near capacity, and engineering can support only a few launches before the seasonal traffic spike.
The VP of Growth wants fast wins and pushes for running more experiments in parallel. The Head of Data Science wants stronger evidence thresholds and is concerned about underpowered tests. Engineering wants to minimize one-off implementation work that creates tech debt. Finance wants confidence that discount-related ideas will not reduce ARPU.