Diagnose a steady three-month rise in phone returns at Flipkart

You are the product manager who owns returns and cancellations at Flipkart, the Indian e-commerce marketplace. Over the last three months, returns in the mobile phone category have been climbing slowly but steadily. Describe what you would do: how you would work out what is behind the increase, and how you would respond once you know. ```hint Count is not rate Before hunting for causes, check whether more returns simply reflect more phones being sold, and remember that a return is filed some days after delivery, so think about which orders each return belongs to. ``` ### Clarifying Questions - Is the rise in the absolute number of returns, or in returns as a share of phones delivered? Have phone sales grown over the same period? - Does the mobile category here mean handsets only, or also accessories such as chargers and cases? - Do replacement requests count as returns, and are orders refused at the doorstep included, or only requests made after delivery? - Is the increase seen only in phones, or in other electronics categories as well? ### What a Strong Answer Covers - Validation of the metric (definition, normalization by deliveries, attribution to delivery date) before searching for causes. - Use of the gradual shape to favor causes that build over time over one-day events. - Segmentation by brand and model, seller and fulfillment type, geography, payment method and delivery route, combined with return reasons and timing, to localize the rise. - Internal and external hypotheses, each paired with the evidence that would confirm it and a matched response. - Success metrics and guardrails covering customer experience, conversion and the cost of returns. ### Follow-up Questions - The rise turns out to come almost entirely from one popular new model sold by many sellers. What do you do next, and whom do you involve? - Tighter checks at return pickup cut fraudulent returns, but satisfaction falls among customers with genuine returns. How do you balance the two? - How would you set up monitoring so the next gradual rise is caught within weeks rather than months?