Personalizing Amazon's Homepage: Shopper Intents, Candidate Ranking and Page Layout

How would you design the engine that personalizes Amazon's homepage, choosing which products each shopper sees when they open it? ```hint Start from why people open the page A shopper restocking coffee, one tracking a parcel and one hunting for a gift want very different things from the same page. Work out the jobs the homepage does for each before deciding how any single product gets picked. ``` ### Clarifying Questions - What should the homepage optimize: purchases in this visit, long-term customer value, or discovery of other Amazon offerings such as Prime Video? - Which shoppers are in scope: signed-in shoppers with history, or also signed-out visitors and brand-new customers? - Are sponsored products placed by this engine, or by a separate ads system that the engine has to make room for? - Should the answer focus on the shopper experience, the underlying system, or both? ### What a Strong Answer Covers - The homepage's goals for shoppers and for Amazon, and the shopper intents the engine must serve. - The signals and the pipeline that narrow a huge catalog down to one personal page, including how the page's sections are chosen and ordered. - Handling of new shoppers and new products, freshness, variety, and items the shopper just bought. - Success metrics tied to incremental value rather than clicks, with guardrails for trust, sponsored content and page speed. - A testing and rollout plan, plus risks such as feedback loops and privacy. ### Follow-up Questions - A shopper buys a refrigerator and the homepage fills with refrigerators. How should the engine tell one-time purchases from items people buy again? - How would you measure the homepage's incremental effect when many shoppers would have bought the same items through search anyway? - Marketing wants a guaranteed top slot for a seasonal campaign. How do you decide?