A customer reports a data issue that only occurs for one account or one configuration. The pipeline is otherwise healthy for other customers, so the investigation needs to isolate account-specific inputs, config, orchestration history, and transformation behavior without assuming a global outage.
Global dashboards may look normal while one account is failing.The bug may come from source data, config drift, or account-specific branching logic.A partial rerun can create duplicates if the pipeline is not idempotent.You need a path to validate the fix and backfill safely.