Debug A Duplicate Fact Model
Diagnose and fix duplicate rows caused by a many-side join in dbt.
fct_orders fails a unique order_id test after payments are joined. Ask why the grain changed, not just why the test failed. The shortcut is to patch the final mart because it is the visible failure. That hides whether the defect is raw data, staging cleanup, join grain, or metric definition. Inspect failing rows Query duplicated order_ids and count payment rows per order. The evidence shows whether duplicates are real orders or join multiplication. Pre-aggregate many side Build int_payments__order_rollup with one row per order before joining. The join now preserves order grain. Prove the contract Keep the unique/not_null test on fct_orders.order_id…
Sign up free — one personalized lesson every day, matched to your role and goals.
Already have an account? Sign in