Facebook Shopping Ads Delivering Not-as-Pictured Goods: Size the Problem and Design Fixes

On Facebook, some people who buy products they discovered through an ad report that the item that arrived was not what the ad showed, or was of far lower quality than the ad suggested. You are a product manager at Meta. How would you decide whether this is a problem worth solving, and what changes to the product would you make to address it? ### Clarifying Questions - Do these purchases finish inside Facebook (an in-app shop or checkout) or on the advertiser's own website after a click? The answer changes what Meta can see and control. - How did the reports surface (in-app reports, press coverage, a regulator, advertiser complaints), and how many reports does "some people" mean? - Is the priority protecting buyers, protecting the value of ads for honest advertisers, or meeting a legal obligation? - Are the reports concentrated in particular product categories or countries? ### Part 1 — Is this problem worth solving? Explain how you would size the problem and decide whether it deserves investment compared with other priorities. ```hint A handful of reports is not a size People who complain are a small, biased slice of all buyers. Think about how you would estimate the true rate across ad-driven purchases, and what a bad purchase costs Meta beyond the one buyer. ``` #### What This Part Should Cover - Ways to measure prevalence beyond complaint counts, and how concentrated the problem is - The impact on buyers, on honest advertisers and on Meta's revenue and trust - A clear decision rule that weighs the size of the problem against the cost of solving it ### Part 2 — Product changes Assuming the problem is worth solving, propose the product changes you would make, choose which to build first, and say how you would measure success. ```hint Follow the purchase journey A mismatch can be stopped before an ad runs, flagged while a person is deciding to buy, or put right after the package arrives. Place each idea on that journey and ask who pays its cost. ``` #### What This Part Should Cover - Solutions that act at different stages of the journey, using signals Meta uniquely has - Prioritization by impact, effort and side effects on legitimate advertisers - Success metrics, guardrails and how you would test the changes ### What a Strong Answer Covers - Separating outright fraud from misleading but genuine listings and from subjective quality disappointment, and handling each differently - Evidence-based sizing that turns user harm into an effect on future ad engagement and revenue - A prioritized set of fixes rather than an unranked list - The balance between buyer protection and false positives, advertiser friction and ad revenue - How bad actors would respond and how the system keeps up with them ### Follow-up Questions - Your changes cut reports sharply, but ad spend from small advertisers also falls. How would you decide whether that trade is acceptable? - Banned sellers keep coming back with new pages and new ad accounts. What would you change? - How would you measure the problem for purchases that complete on outside websites Meta cannot observe?