Diagnose a 30% spike in mis-shipped orders at an online fashion retailer

You are a product manager at an online retailer that sells fashion and lifestyle products. Wrong deliveries have increased by 30%. Explain how you would track down the root cause and what you would do to fix the problem quickly. ```hint Split the error types first A different product, the right product in the wrong size or color, a missing item and a parcel left at the wrong address break at different points between the order and the doorstep. Work out which kind of error grew before deciding where to look. ``` ### Clarifying Questions - Which errors count as a wrong delivery here: a different product, the right product in the wrong size or color, a missing item, or a parcel left at the wrong address? - Is the 30% a rise in the rate of wrong deliveries or in their raw count, over what period, and did it appear suddenly or build gradually? - Where does the number come from: customer complaints, the reason chosen when requesting a return or exchange, or inspection of returned items? - Is the rise spread evenly, or concentrated in particular fulfillment centers, regions, categories, sellers or carriers? ### What a Strong Answer Covers - A precise definition of the metric and a check that the 30% reflects real errors rather than a change in tracking, reporting or order volume - Segmentation of the rise by error type, site, category, supplier, platform and time to localize it - A map of the order's path from product listing to doorstep, with a testable hypothesis for each step, checked in a sensible order - Immediate containment for affected customers and in-flight orders, alongside a permanent fix for the failing step - Metrics that confirm the fix, with guardrails on fulfillment speed, cost and false claims ### Follow-up Questions - Returned items match their pick lists but not the orders customers placed. What does that tell you, and where do you look next? - The quickest safeguard is a second scan of every item at packing, which slows each order. How do you decide whether and where to use it? - How would you tell genuine wrong deliveries apart from customers who report a wrong item to make a return free?