Erase people from the warehouse within the month

How to do it

Drive warehouse deletion from the same erasure log the app uses, by the person's ID and pseudonym. Delete in raw first, then rebuild or delete in every later table, found through Record lineage from the code. Warehouses keep deleted rows for a while, such as up to 7 days of time travel plus 7 days of fail-safe in BigQuery. Count that time inside the month. Spreadsheet exports and BI caches count too. Anonymising instead of deleting works only when nobody can single the person out again, as Delete old data automatically explains. The EDPB's 2025 review of erasure found weak anonymisation used in place of deletion, and deletion from backups handled badly.

Examples

Python example

def erase_from_warehouse(warehouse, person_id: str, person_pseudonym: str, tables: list[str]):
    with warehouse.transaction():
        for table in tables:
            warehouse.execute(f"DELETE FROM {table} WHERE person_id = %s OR person_key = %s",
                              (person_id, person_pseudonym))
    erasure_log.mark_done(person_id, store="warehouse")

JavaScript example

export async function eraseFromWarehouse(warehouse, personId, personPseudonym, tables) {
  await warehouse.transaction(async tx => {
    for (const table of tables) {
      await tx.query(`DELETE FROM ${table} WHERE person_id = $1 OR person_key = $2`, [personId, personPseudonym]);
    }
  });
  await erasureLog.markDone(personId, { store: 'warehouse' });
}

How to check

Erase a test person with rows in every source. Within 30 days, a search of every warehouse table, extract and BI cache finds no trace of their ID, pseudonym or email. Time travel cannot bring them back.

Sources

• GDPR Articles 12, 17 and 19
• EDPB report on the 2025 coordinated action on the right to erasure

Related rules

• DEN-9 Carry deletes into the warehouse in Data engineering
• DEN-19 Record lineage from the code in Data engineering
• WEB-217 Let people see and delete their data in Website
• DAT-31 Delete old data automatically in Data and databases