Success Metrics and Experiment Design for a New Uber Eats Promotions Tab

Uber Eats has added a new promotions tab to its app. As the product manager responsible for it, how would you decide whether it is working? ```hint Incremental, not just used A busy tab is not necessarily a successful one. Ask what the same customers would have ordered, and what Uber would have spent on discounts, if the tab did not exist. ``` ### Clarifying Questions - What is the tab meant to achieve: more frequent orders from existing customers, winning back lapsed ones, first orders from new customers, or more sales for merchants running their own offers? - Who funds the discounts shown in the tab: Uber, the merchants or both? - Has the tab already rolled out to everyone, or can a group of users still be held back from seeing it? - Does it show only restaurant deals, or also grocery and convenience offers and membership perks? ### What a Strong Answer Covers - A stated goal for the tab, covering both the eaters and the merchants it serves - One primary success metric tied to incremental orders or gross bookings rather than tab visits - A funnel for the tab itself, from opening it to placing an order, to explain why the primary metric moved - Guardrails for discount cost, cannibalization of full-price orders, discovery elsewhere in the app, delivery reliability and merchant outcomes - An experiment or holdout that isolates the tab's effect, with a plan for duration, novelty and long-term discount dependence ### Follow-up Questions - Orders placed through the tab are up, but orders per customer are flat overall. What does that tell you, and what would you do? - How would you judge whether the tab is a success for merchants? - How would you detect customers learning to wait for discounts, and what would you change if you saw it?