Diagnosing and Reversing a 10% Drop in App Conversion at Uber
Company: Uber
Role: Product Manager
Category: Analytics & Experimentation
Difficulty: medium
Interview Round: Other
You are a product manager at Uber, and you notice that an app's conversion rate has dropped by 10%. How would you investigate the cause, and what would you do to resolve it?
```hint Split the ratio
A conversion rate falls either because fewer people complete the step or because more people enter it. Decide which side moved, and at which step of the funnel, before you name causes.
```
### Clarifying Questions
- Which app and which conversion: rider sessions that end in a trip request, new installs that reach a first trip, driver sign-ups that complete onboarding, or Uber Eats visits that end in an order?
- Is the 10% a relative decline (for example, 30% falling to 27%) or a drop of 10 percentage points?
- Did the drop appear on a single day or build over several weeks, and does it show on both iOS and Android and in every city?
- Can I see the release, pricing and experiment logs for the period when the drop started?
### What a Strong Answer Covers
- Pinning down the conversion definition, its numerator and denominator, and the size and timing of the drop
- Ruling out tracking, definition and traffic-mix artifacts before treating the drop as real
- Localizing the decline to one funnel step and to segments such as platform, app version, city and new versus returning riders
- Ranked hypotheses for a two-sided marketplace (releases, pricing, wait times, driver supply, payments, competitors, seasonality), each with the check that would confirm it
- A resolution plan: mitigation matched to the cause, stakeholder updates, recovery metrics with guardrails, and alerting
### Follow-up Questions
- The drop sits entirely between seeing the fare and tapping Request, and only in a handful of cities. What do you check first, and which fixes would you weigh?
- Sessions jumped after a marketing push while trip requests stayed flat. Is that a problem to fix, and how would you explain it to leadership?
- How would you set up monitoring so a drop like this is caught within a day instead of at a monthly review?
Overview: A product execution question for an Uber PM interview: an app's conversion rate has fallen by 10%, and the candidate must explain how to find the cause and fix it. It tests defining the funnel metric, ruling out data issues, segmenting to localize the drop, weighing marketplace hypotheses and planning the recovery.