Uber Daily Trips Fall by Half: Finding the Root Cause of a Metric Drop
Company: Google
Role: Product Manager
Category: Analytics & Experimentation
Difficulty: medium
Interview Round: Other
You are a product manager on Uber's ride-hailing product. Uber's daily ride count has fallen to half its usual level. Work out why: explain how you would track down the cause, which data you would look at and in what order, and how you would confirm the root cause.
```hint Confirm, then localize
Before looking for causes in the outside world, make sure the drop is real and pin down its shape: which definition of a ride fell, how quickly, and where. Then split the ride funnel to see which step broke.
```
### Clarifying Questions
- Does "rides" mean ride requests, matched trips or completed trips, and is the 50% measured against the previous day, the same weekday in earlier weeks, or last year?
- Did the drop happen overnight or build up over weeks?
- Is it global, or concentrated in certain cities, countries, ride products or app platforms?
- Are Uber's other businesses, such as food delivery, down over the same period?
### What a Strong Answer Covers
- Clarifying the metric and the drop's shape (timing, scope, comparison baseline) before forming hypotheses
- Ruling out a data or logging problem first, with a concrete way to cross-check the number
- A structured set of hypotheses spanning internal causes (releases, outages, pricing, incentives) and external ones (competitors, events, regulation, platform outages), on both the rider and the driver side
- Using funnel steps and segment cuts to localize the drop, checked in a sensible order by likelihood and cost of checking
- Confirming that the root cause explains the timing, scope and size of the drop, and the next steps for each kind of cause
### Follow-up Questions
- App opens and ride requests look normal, but completed trips have halved. Where do you look next?
- Suppose the drop is confined to one city where a competitor just launched a large discount. How would you respond, and what would you avoid doing?
- How would you change monitoring so a drop this large is caught within hours instead of in the next daily report?
Overview: A metric diagnosis question about Uber's daily ride count falling to half its usual level. It tests clarifying the metric and baseline, ruling out data errors, segmenting the drop, splitting the trip funnel across riders and drivers, and ordering internal and external hypotheses to confirm a root cause.