Uber fare splitting: north-star metric, funnel, guardrails and how to measure them
Company: Uber
Role: Product Manager
Category: Analytics & Experimentation
Difficulty: medium
Interview Round: Other
Uber's rider app lets people who share a ride split what it costs. Assume the feature works as it is commonly described: during the trip, the rider who booked it invites the others in the car, each invitee accepts or declines in the app, and the fare is divided among everyone who accepts, with each person charged on their own payment method. You are the product manager who owns this fare-splitting feature. What success metrics would you define for it, and how would you measure them?
```hint Follow one split
A split only succeeds if other people respond and pay. Trace a single split from the requester's first tap to the last payment, and note every place it can stall.
```
### Clarifying Questions
- Is the main purpose of the feature to win more group trips, to bring new riders onto the platform through invitations, or to remove an awkward payment moment for people who already ride?
- Is this a recent launch where adoption matters most, or an established feature whose health and incremental value are in question?
- When an invitee declines or their payment fails, does the requester pay the remainder, or does Uber absorb it?
- Must invitees already have an account, or can they sign up from the invitation?
### What a Strong Answer Covers
- A goal for the feature from which the headline metric clearly follows, rather than metrics listed without a purpose.
- A step-by-step funnel from seeing the option to every share being paid, with a metric at each step.
- Measures of incremental value (extra trips, new riders acquired through invitations, retention) separated from what riders would have done anyway.
- Guardrails such as failed or uncollected shares, support contacts about charges, and effects on drivers.
- A measurement plan that deals with self-selection and with the fact that one split involves several riders.
### Follow-up Questions
- Split usage doubled after the button moved to the in-trip screen, but support contacts about charges also rose. How do you decide whether to keep the change?
- How would you design an experiment when a requester and their invitees could fall into different test groups?
- Riders who split fares take far more trips than riders who never do. Why can't you conclude the feature caused that, and what would you do instead?
Overview: Define the success metrics for Uber's in-app fare-splitting feature and explain how each would be measured. It tests goal setting, funnel thinking from invitation to payment, guardrails, and experiment design that copes with self-selection and with riders who may land in different test groups.