Diagnosing a 10% three-month decline in Uber Eats cart-to-order conversion
Company: PayPal
Role: Product Manager
Category: Analytics & Experimentation
Difficulty: medium
Interview Round: Other
You are the PM for Uber Eats. Cart conversion has fallen 10% over the past three months. How would you work out what is causing it?
```hint Let the shape guide you
A steady slide and an overnight cliff point to different causes. Before listing hypotheses, think about what the trend line, the count of orders and the count of carts would each tell you on their own.
```
### Clarifying Questions
- How is cart conversion defined: orders placed divided by carts created, by sessions with an item in the cart, or by users, and does an order whose payment fails count?
- Is the drop 10% relative (for example from 50% to 45%) or 10 percentage points?
- Did the decline start on a particular date and then hold, or has it slid steadily across the whole three months?
- Which changes do we already know about in that window, such as app releases, fee or menu price changes, or changes to promotions and the membership program?
### What a Strong Answer Covers
- Confirming the drop is real: the metric's definition, logging and app-version changes, and an independent count of orders.
- Splitting the ratio into orders and carts, and the path from cart to placed order into steps, to find where users leave.
- Segment cuts that tell a mix shift apart from a decline within segments.
- Hypotheses across eaters, merchants, couriers, payments and outside factors, ranked by how well they fit the observed shape and timing of the decline (a step on one date or a steady slide), each with a way to test it.
- Sizing how much each confirmed cause explains, choosing the fix, and the metrics to watch afterward.
### Follow-up Questions
- Suppose you find the decline is concentrated in carts whose fees rose during the window and are shown only at checkout. How would you decide between changing the fees and changing how they are shown?
- Conversion fell but total orders are flat. Is that a problem, and what would you tell leadership?
- How would you set up monitoring so that a decline like this is flagged within days or weeks instead of being noticed three months in?
Overview: A product analytics case where Uber Eats cart conversion has fallen 10% over three months and the PM must find out why. It tests validating the metric, breaking down the checkout funnel, separating mix shifts from real declines and ranking hypotheses across eaters, couriers, merchants and payments.