The previous version of the data is gone. We have a new version of the data, but it is incorrect. We cannot give it to downstream systems or the business.

When a new data version is incorrect, first isolate the issue by running validation rules (schema checks, null counts, range checks). Use data lineage tools to trace the source of the error. If the previous version is lost, implement a rollback strategy: keep a snapshot of the last good state in a separate bucket or table. Once the issue is fixed, re‑load the corrected data and run a full audit against downstream consumers. Communicate the delay to stakeholders and provide a rollback plan if the new data is already in use. Establish a data quality dashboard and automated alerts to prevent future incidents.