How to set up test data for automated runs?

Define what each test needs before it runs: user accounts, orders, subscriptions, or records in a known state. Avoid shared “magic” accounts that other tests modify unless you reset state after every run. Create data through APIs or setup scripts when possible - faster and more reliable than clicking through long UI flows to seed the same state. Use synthetic or anonymized data in lower environments; do not point automation at unpredictable production copies. Clean up or isolate per run: unique emails, transactions, or database snapshots so parallel jobs do not collide. Document dependencies in the suite README or test metadata. Good data setup is what makes automated runs repeatable in local dev and CI at scale.