Decide Whether to Add Shoppable Product Recommendations to the Facebook Feed

Meta is weighing a new unit in the Facebook News Feed that recommends products a person might want to buy. You are the product manager. How would you decide whether to build it? Explain what you would need to believe about users, sellers and the business, how you would gather that evidence, and what result would lead you to build it, reshape it or drop it. ```hint Compete for the slot Every unit in the feed takes space from friends' posts, groups and ads. Ask what this unit must deliver, for people and for Meta, to beat whatever it would push down, and how you could learn that cheaply before a full build. ``` ### Clarifying Questions - Where would the recommended products come from: Marketplace listings from individuals, catalogs that businesses upload, paid placements from advertisers, or a mix? - Would people buy without leaving Facebook, or would tapping a product open the seller's listing or website? - Is the main aim helping people discover things they want, opening a new revenue stream, or responding to shopping features in competing apps? - Is the unit meant for every user, or only for people who already show interest in shopping on Facebook? ### What a Strong Answer Covers - Why Meta would want a shopping unit in the feed, tied to its mission and to its ads and commerce business, including the cost of the feed space it would take. - The groups involved (shoppers, casual users, individual sellers, businesses, advertisers), their pain points, and a reasoned choice of whom to serve first. - The assumptions that must hold for the unit to be worth building, each paired with evidence or a test and a go/no-go threshold set in advance. - A primary success metric based on incremental purchases or seller contacts, with guardrails for feed engagement, ad revenue, hides and reports, and user trust. - Risks such as feed clutter, recommendations that feel invasive, scams and low-quality sellers, and how the plan would limit them. ### Follow-up Questions - The test shows a high click rate, but people hide the unit far more often than they hide ads. Do you launch it? - Purchases rise, yet total ad revenue falls because the unit takes slots that used to show ads. How do you make the call? - How would you keep recommendations based on someone's activity from feeling intrusive? - Suppose the unit is instead a module called "Products you might want to buy" inside the Marketplace feed, which already shows suggested listings. How would your goal, success metric and test change, and which of the two surfaces would you try first?