Self-assessment

Score yourself

Answer each question honestly. Yes = 2, Partly = 1, No = 0. Keep a running total.

Your score

0/ 24

0 of 12 answered

Not scored yet

The schedule

0 / 6

01 Is there a written retention schedule: one row per record class, with the period and the purpose or law behind it?

02 Was it built from your [data inventory](/enterprises/focus-areas/data-discovery-classification-and-mapping), not from memory?

03 Where two duties collide, one says keep, one says delete, is the resolution recorded, with its legal basis?

The deletion

0 / 6

04 When a period ends or consent dies, does deletion actually run, on schedule, not on intention?

05 Does erasure reach live systems, archives and backups, including copies at processors?

06 Does "delete" mean delete? If the record still exists but is hidden from a screen, it has not been erased.

The location

0 / 6

07 Do you know which of your datasets must stay in India under sectoral rules?

08 For everything else, is the hosting posture a recorded decision, with reasoning you could show a regulator?

09 Can anyone reach production from outside India, and is that access authorised in writing and logged?

The proof

0 / 6

10 Is there proof of each deletion run: what was deleted, from where, when, under which rule?

11 Is your backup handling stated honestly: deletion in live systems and archives, plus a restore-and-re-delete routine so a restored backup does not silently bring deleted data back?

12 Pick a record class whose retention period ended last quarter: can you produce the proof that deletion ran everywhere, including at processors?

Readiness bands for this self-assessment
Score Where you stand What to do next
0 to 8ExposedData is kept forever by default. Build the retention schedule from the inventory this quarter.
9 to 16HalfwayThe schedule exists but execution leaks: deletion jobs, backup handling or processor copies.
17 to 21Strong baseProve it with the deletion test below, then automate the runs with logs.
22 to 24LeadingKeep the schedule current as new systems and new laws arrive; review at least annually and on every major change.

Any single "No" is a named gap, whatever the total says. Nothing leaves your browser.

What the law requires

Personal data must be erased when the person withdraws consent or the purpose it was collected for is no longer being served, unless the law requires you to keep it. That is a legal duty. Tax rules, KYC directions and sectoral regulations often force fixed retention periods, and those duties override deletion.

On localisation, be precise: the DPDP Act does not generally force personal data to stay in India. It restricts transfers to countries the Government notifies as off-limits. Do not assume "DPDP requires India hosting"; it does not. Some sectoral rules do require India storage for specific data, the payments data directions from the banking regulator are the standard example, and many regulated customers insist on India-only hosting in their contracts. The data sovereignty page goes deeper.

The framework: four moves

Move 1: Build the schedule from the inventory. One row per record class. Purpose first, then legal minimums; record which law sets each floor. Where a retention duty and an erasure duty collide, record which law wins and why.

Move 2: Execute erasure with proof. Write the deletion method per class: live systems, archives, backups, who runs it, how often. Automate the runs; deletion jobs with logs beat calendar reminders with good intentions. Keep the proof of each run. And be honest about backups: instant deletion from every backup is not how backups work, so state the restore-and-re-delete routine instead.

Move 3: Resolve the conflicts on paper. A conflict register: where duties collide, the resolution, its legal basis, and who decided. Legal holds override deletion; record who can place one.

Move 4: Decide the localisation posture deliberately. Map which datasets fall under sectoral India-storage rules. For the rest, choose the hosting posture as a policy decision and record the reasoning. If your contracts with banks or government bodies demand India-only storage and no remote production access from outside India without written authorisation, the posture must match.

Template: the retention schedule

One row per record class. Copy this and fill it in.

Record classPurposeKept untilThe law behind any minimumDeletion methodOwnerLast runProof kept where
Example: loan filesService the loan3 years after closureSectoral record rulesPurge job, live + archive; restore-and-re-delete on backupsLending opsDeletion log store
Example: marketing consentsMarketing while consentedUntil withdrawalNoneImmediate suppression + purgeMarketing opsDeletion log store

The downloadable file adds the conflict register and a column guide.

What good looks like at the end

You should have: a retention schedule with legal bases, deletion procedures that actually run, proof of deletion runs, the backup handling stated honestly, and a recorded localisation posture with contract terms to match.

The live test: pick a record class whose retention period ended last quarter. Produce the proof that deletion ran, across live systems, archives and backups, including at processors holding copies.

Where enterprises go wrong

  • Retention rules written but never executed.
  • "Delete" that means "hide".
  • No record that deletion happened, so compliance cannot be shown.
  • Promising instant deletion from backups, then failing to deliver it.
  • Assuming the DPDP Act forces India hosting, or assuming it forbids nothing. Neither is true.

The decisions to make

  • Erasure depth: live systems only, or archives and backups too.
  • How legal holds override deletion, and who can place one.
  • The hosting posture per data class, and the reasoning you will show a regulator.
  • How sectoral localisation rules stricter than the general law are tracked.

Next step

Security controls that make protection real, the safeguards the data you keep still needs. If you need help with any of this, write to us and we will help.