Use data to resolve an ambiguous problem

Tell me about a time you used data to solve a poorly defined business problem end-to-end. Specify the exact hypothesis, the data sources and their known biases, your statistical method or model selection, and how you validated assumptions. Quantify impact with a defensible counterfactual (e.g., difference-in-differences vs. naive before–after). Include how you handled data quality issues, communicated uncertainty to stakeholders, and one thing you would improve if you had 10% more data.