Amber Grid: Trial-to-paid conversion Dropped 8%
Product Management
easy40 min0 submissionsFlipkart
Scenario
You are the PM for Amber Grid's core utilities product.
Trial-to-paid conversion has fallen 8% over the past 18 days. The drop appears concentrated in Android and is most pronounced in the US.
Context that may or may not be relevant:
- A redesigned onboarding flow shipped 5 days ago to 31% of users
- Marketing paused a paid acquisition campaign 16 days ago
- A competitor launched a free tier last month
- The data team migrated the analytics pipeline 17 days ago
Your VP wants an answer by end of day and is already asking whether the redesign should be rolled back.
Supporting data
incident
- metric
- trial-to-paid conversion
- drop pct
- 8
- window days
- 18
- concentrated region
- the US
- concentrated platform
- Android
recent changes
- onboarding rollout pct
- 31
- onboarding redesign days ago
- 5
- competitor free tier launched
- last month
- paid campaign paused days ago
- 16
- analytics pipeline migrated days ago
- 17
Your task
Diagnose the drop. Provide:
- Analysis — how you would structure the investigation and which hypotheses you would test in what order.
- Risks — of acting too early, and of acting too late.
- Recommendation — your diagnostic sequence and what you would tell the VP today.
Be specific about the data cuts you would pull.
Ready to move forward? Up next: Market Sizing Sprint: India ConsumerNext question
How you'll be graded
100 points, 60% to pass.
- data analysis25
- recommendation25
- problem structuring25
- hypothesis generation25
Hint
Reveal suggested structure
- Is it real? Check instrumentation first — the pipeline migration is the cheapest hypothesis to eliminate and the most embarrassing to miss.
- Decompose the metric. For DAU: new + returning + resurrected − churned. Which component moved?
- Segment. Platform, region, cohort, acquisition channel, new vs existing users.
- Timeline. Does the drop align with a specific deploy, or is it gradual?
- Internal vs external. Internal changes are testable and reversible; external ones are not.
- Act on the evidence, with a rollback only if the evidence points there.