Who This Helps
You're a team lead who wants to scale a repeatable analytics routine. You have a list of possible experiments, but you're not sure which one to run next. This article from the Product Portfolio Strategy course helps you prioritize so you don't waste time on low-impact ideas.
Mini Case
Imagine your team has three experiment candidates: A (improve onboarding), B (add a referral feature), and C (optimize checkout). You have rough data: A could boost activation by 12%, B might increase sign-ups by 8%, and C could lift revenue by 5%. But you only have capacity for one experiment this sprint. Using bet sizing from the course, you estimate confidence and effort. A has high confidence (80%) and low effort (3 days), B has medium confidence (50%) and medium effort (5 days), C has low confidence (30%) and high effort (7 days). You pick A. The result? Activation jumps 12% in two weeks. That's a win.
Do This Now (5 Steps)
- List all experiments your team is considering. Keep it to 5 or fewer.
- For each, estimate the potential impact (e.g., 12% improvement) and your confidence (e.g., 80%).
- Estimate effort in days or hours. Be honest, not optimistic.
- Rank experiments by impact times confidence divided by effort. Highest score wins.
- Pick the top experiment and assign it to a team member. Start tomorrow.
Avoid These Traps
- Don't pick an experiment just because it's fun. Use the numbers.
- Don't skip confidence estimation. Low confidence means high risk.
- Don't forget to check if the experiment aligns with your portfolio guardrails (like not degrading customer support response time).
- Don't overthink it. A quick bet sizing is better than no sizing.
- Don't ignore capacity. If your team is already overloaded, pick a low-effort experiment.
Your Win by Friday
By Friday, you'll have a clear #1 experiment to run. Your team will know exactly what to work on. You'll avoid the chaos of random prioritization. And you'll have a repeatable routine for next time. Plus, you'll feel like a prioritization ninja. Not bad for a week's work.