Effective risk systems begin with disciplined data foundations. Every registration, clickstream event, listing action, order, shipment update, payment attempt, and dispute produces traces that can be transformed into risk signals. This chapter maps the e-commerce data ecosystem across buyers, sellers, and platform infrastructure, covering identity attributes, device and network telemetry, behavioral sequences, transactional artifacts, fulfillment records, support interactions, and external intelligence. Beyond cataloguing sources, it emphasizes reliability: Time-respecting datasets, leakage prevention, stable feature semantics, and label quality are prerequisites for valid modeling. Fragmented logging or inconsistent definitions can undermine even sophisticated algorithms. The chapter also addresses governance fundamentals-data lineage, controlled access, documentation standards, and privacy compliance-which enable reproducibility and defensibility in high-stakes decisioning. We also discuss practical data engineering choices-aggregation windows, entity resolution, and feature freshness-that determine whether models can score accurately in real time and remain stable under product change. The aim is to make the data layer explicit so later modeling chapters can be read as design decisions rather than abstract techniques.