Cobalt Robotics: Users Sign Up and Disappear
Product Management
medium45 min0 submissionsAmazon
Scenario
Cobalt Robotics's industrial robotics product acquires users steadily but struggles to keep them.
Retention curve (all users):
| Day | Retained |
|---|---|
| D1 | 66% |
| D7 | 27% |
| D30 | 20% |
| D90 | 9% |
What else we know:
- 30% of new users complete the core action at least once in week one
- Among users who complete the core action three or more times in week one, D30 retention is 60%
- 18% of users account for roughly 76% of total usage
- Exit surveys most commonly say "didn't find it useful" and "forgot about it"
Growth is currently offsetting churn with paid acquisition, at a blended CAC that has risen 49% year on year.
Supporting data
segments
- power users pct
- 18
- activation rate pct
- 30
- power user share of usage pct
- 76
- d30 retention for 3x core action pct
- 60
economics
- cac increase yoy pct
- 49
qualitative
- top exit reasons
- didn't find it useful,forgot about it
retention curve
- d1
- 66
- d7
- 27
- d30
- 20
- d90
- 9
Your task
Diagnose the retention problem and propose a plan. Provide:
- Analysis — what the curve and the segment data tell you.
- Risks — of the interventions you propose.
- Recommendation — what you would build, and how you would measure success.
Ready to move forward? Up next: Market Sizing Sprint: India ConsumerNext question
How you'll be graded
100 points, 60% to pass.
- root cause25
- measurement25
- cohort analysis25
- solution design25
Hint
Reveal suggested structure
- Read the curve's shape. A steep D1→D7 fall is an onboarding or value-discovery problem. A curve that never flattens is a product-market fit problem. These need opposite responses.
- Find the habit moment — the behaviour that separates retained from churned users.
- Compare segments — the 3×-core-action cohort is the natural experiment already in the data.
- Attack the biggest drop-off first.
- Instrument the intervention so the effect is measurable.