Short answer: a DPDP rights-request workflow is one pipeline that takes every access, correction, erasure, grievance and nominee request, checks who is asking using identifiers you already hold, and fans the request out to every system and vendor that holds the person's data. Each request moves through fixed states (received, verified, in progress, partial, done, or rejected with a reason) and every step is logged. Where a law requires you to keep data, as can be the case for KYC or tax records, you restrict it and record why instead of deleting it. Rule 14 requires you to publish how to make a request and to answer grievances within a period you publish, which cannot exceed 90 days. These duties start on 13 May 2027.
Our DPDP implementation guide for engineering teams sets out the five components a DPDP build needs. This post goes deep on one of them: the request router and the workflow around it. Legal wording for notices and responses should come from your counsel.
Which DPDP rights must your systems support, and what does each mean?
The Act gives Data Principals four rights: access (Section 11), correction, completion, updating and erasure (Section 12), grievance redressal (Section 13) and nomination (Section 14). Section 8(7) adds erasure when consent is withdrawn or the purpose is no longer being served, unless a law requires retention, and requires you to make your processors erase too. Each right becomes a different job in each system.
| Right | What the person gets | System actions |
|---|---|---|
| Access (s.11) | What the person gets: Summary of data and processing; list of Fiduciaries and Processors it was shared with | System actions: Query each system by person ID; pull categories and purposes, not raw dumps; join with the vendor register; apply the s.11(2) exception to the sharing list |
| Correction, completion, update (s.12(2)) | What the person gets: Accurate, complete, current data | System actions: Update the source of truth; push the change to copies in CRM, warehouse and support tool; record old and new value hashes, not the values |
| Erasure (s.12(3)) | What the person gets: Data erased unless needed for the purpose or required by law | System actions: Delete or anonymise in each system; where a law needs it, move to a restricted store with a hold reason and purge date; tell processors to erase |
| Withdrawal of consent (s.6, s.8(7)) | What the person gets: Processing stops; data erased unless law requires retention | System actions: Stop the purpose first, then run the erasure path for data tied only to that purpose |
| Grievance (s.13, Rule 14(3)) | What the person gets: A response within your published period, at most 90 days | System actions: Ticket with a due date; link to any related request; route to the owner of your grievance mechanism (s.8(10)) |
| Nomination (s.14, Rule 14(4)) | What the person gets: A named nominee can act on death or incapacity | System actions: Store nominee details on the account; accept nominee requests after a trigger check; act with the same scope as the principal |
Two scope points matter. Sections 11 and 12 attach to data processed on consent, including data given voluntarily under Section 7(a), so tag each data set with its legal basis and ask counsel where the line falls. And Section 11(2) takes one kind of sharing out of the sharing list: data given to another Data Fiduciary that is authorised by law to obtain it and asked in writing, for preventing, detecting or investigating offences or cyber incidents, or for prosecuting or punishing offences. Your access report needs a way to flag those disclosures and leave them out. The summary of your own processing under Section 11(1)(a) is still due.
What does Rule 14 make you publish?
Rule 14(1) says you must prominently publish on your website or app the means for making a rights request, and the username or other identifier needed to identify the person under your terms of service. Rule 14(5) defines identifier broadly: a customer ID, application reference number, email address, mobile number or licence number, for example.
Rule 14(3) requires you to publish the period within which you respond to grievances. That period must be reasonable and cannot exceed 90 days. You also need technical and organisational measures to meet it. Rule 9, which gives effect to Section 8(9), adds the business contact details of your Data Protection Officer, if you have one, or of a person who can answer questions about processing. You publish them prominently and mention them in every response to a rights communication. Section 8(10) separately requires an effective mechanism to redress grievances, and Section 13(3) says the person must use it before going to the Board.
The Act and Rules set no fixed deadline for access, correction or erasure, only the 90-day outer limit for grievances. An unanswered request soon becomes a grievance, so set your own internal target for each request type well inside 90 days. Ask counsel what period to publish.
How do you verify identity without collecting more data?
The safest identity check uses what you already hold. Asking for an Aadhaar copy or a selfie to process a deletion request creates new personal data you then have to protect and erase. Rule 14(1)(b) points the other way: publish which identifier you need, then match against it.
- Logged-in session. A request from inside the app is already tied to the account. Make this the main path.
- One-time code to the registered channel. For web or email requests, send a code to the mobile or email already on file, never to one supplied in the request.
- Step-up for high-impact actions. Erasing a funded account may justify a second factor you already use for login or payments.
- Nominees are the exception. The Act and Rules do not list what a nominee must show, but you will need some evidence of death or incapacity. Keep those documents in the request's evidence store with a deletion date.
Record the method and result on the request, not the code. After repeated failures, reject with the reason "identity not verified" and explain how to try again. Section 15 puts duties on Data Principals, such as not impersonating another person, not filing false or frivolous grievances, and giving only verifiably authentic information when asking for correction or erasure. Those duties do not replace the check.

Which intake channels should a rights request use?
Make every channel feed one request record.
- In-app privacy page. A form for each right, prefilled from the session.
- Web form. For people without app access, with code verification.
- Grievance and privacy contact address. The published grievance channel and the Rule 9 contact, each with a trigger that creates a request record.
- Support tickets and chat. A tag creates the record. Agents should not delete data by hand in the admin panel.
- Consent Manager. Optional; if you integrate one, requests arrive by API.
Every channel writes the same fields: channel, request type, the identifier given, the time received and the raw text. Section 13(2) counts the grievance period from the date of receipt, so start every clock at receipt, not at verification.
How does a request router reach every system you run?
The router needs two inputs: a register of systems that hold personal data, and a way to find the person in each one. Both come from your data map. Each entry records the owner, identifiers, purposes, legal basis, any retention hold and the connector type.
Connectors come in three types. An API connector calls the system's own delete, export or update endpoint. A job connector runs a script against a database or warehouse. A manual task goes to a named owner with a due date when there is no usable API. Untracked manual work is where requests get lost.
| System | How to find the person | Typical erasure action | Watch for |
|---|---|---|---|
| Primary database | How to find the person: User ID | Typical erasure action: Delete or anonymise rows; cascade to child tables | Watch for: Foreign keys to ledgers you must keep |
| Data warehouse | How to find the person: Hashed user ID or email | Typical erasure action: Delete job across raw and modelled tables | Watch for: Copies in staging tables and old snapshots |
| CRM | How to find the person: Email, phone, CRM contact ID | Typical erasure action: Vendor delete API, not just unsubscribe | Watch for: Duplicate contacts with different emails |
| Support tool | How to find the person: Email, ticket requester ID | Typical erasure action: Redact or delete requester and attachments | Watch for: Screenshots and files inside old tickets |
| Email and messaging | How to find the person: Email, phone | Typical erasure action: Remove from lists and suppress | Watch for: Keep a hashed suppression entry |
| Product analytics | How to find the person: Analytics user ID, device ID | Typical erasure action: Vendor user deletion API | Watch for: Events keyed only by device ID |
| Logs | How to find the person: User ID in log fields | Typical erasure action: Usually retain under Rule 8(3) and Rule 6(1)(e); restrict access | Watch for: Personal data in free-text log lines |
| Backups | How to find the person: Not searchable | Typical erasure action: Let expire on schedule; reapply erasure on restore | Watch for: Long retention snapshots |
| Processors and vendors | How to find the person: Whatever ID you shared | Typical erasure action: Instruction under contract; wait for confirmation | Watch for: Their sub-processors |
| AI tools | How to find the person: Workspace user, uploaded files | Typical erasure action: Delete chats, files and vector entries | Watch for: Embeddings and fine-tuning sets |
AI tools need their own line. A retrieval index built from support tickets still holds the text after the ticket is deleted, so delete the matching chunks too.
Run the fan-out as one sub-task per system, each with its own status.
What states should a rights request move through?
Keep it small and strict. Each transition writes an audit event.
| State | Meaning | Moves to |
|---|---|---|
| Received | Meaning: Logged from any channel; clock running | Moves to: Verified, Rejected |
| Verified | Meaning: Identity confirmed; scope agreed | Moves to: In progress |
| In progress | Meaning: Sub-tasks sent to systems and vendors | Moves to: Partial, Done |
| Partial | Meaning: Some systems pending, or data held under a law | Moves to: Done, In progress |
| Done | Meaning: Every sub-task closed; response sent to the person | Moves to: Closed |
| Rejected with reason | Meaning: Not verified, out of scope or duplicate; reason sent | Moves to: Received, on a retry |
Two rules keep this honest. No request reaches Done while a sub-task is open. And every rejection carries a reason code from a fixed list plus the text sent to the person.
What fields should a request record hold?
A sample shape to adapt. Note what it leaves out: exported data, verification codes and document images.
| Field | Example | Notes |
|---|---|---|
| request_id | Example: RR-2027-000481 | Notes: Used in every response |
| type | Example: access | correction | erasure | withdrawal | grievance | nomination | Notes: One type per record |
| channel | Example: app | web | email | support | consent_manager | Notes: Where it arrived |
| received_at | Example: Timestamp | Notes: Clock starts here |
| due_at | Example: Timestamp | Notes: Internal target |
| subject_ref | Example: Internal person ID | Notes: Not raw email or phone |
| requester_role | Example: principal | nominee | guardian | Notes: Extra checks for nominees |
| verification | Example: method, result, verified_at | Notes: No codes stored |
| state | Example: received ... done | rejected | Notes: From the state machine |
| subtasks | Example: system, connector, status, sent_at, confirmed_at | Notes: One per system and vendor |
| holds | Example: system, data category, legal basis, purge_after | Notes: Why data was kept |
| rejection_reason | Example: Code plus text sent | Notes: Only when rejected |
| response | Example: sent_at, channel, template version | Notes: What the person was told |
| linked_requests | Example: IDs | Notes: Grievance about an erasure |

What happens when erasure conflicts with KYC, tax or Rule 8(3)?
Section 12(3) and Section 8(7) both carve out data you must keep to comply with a law. The Act's own illustration is a bank that must keep identity records after an account closes. In India the pattern is common: KYC records under the Prevention of Money-laundering Act and RBI's KYC directions, books and invoices under tax and GST law, and Rule 8(3), which says a Data Fiduciary shall retain personal data, associated traffic data and other logs of processing for at least one year from the date of processing, for the purposes in the Seventh Schedule. Ask counsel to confirm which of these apply to you and the exact periods for your sector.
The pattern that works is restrict and retain with reason:
- Erase everything that no law or rule requires you to keep. Counsel decides where that line falls, including how far Rule 8(3) reaches.
- Move held data out of live use. A restricted schema or bucket, readable only by named compliance roles, not by product, analytics or AI pipelines.
- Write a hold record. System, data category, the law or rule relied on, and a purge-after date.
- Schedule the purge. A job deletes the data when the hold ends, and writes an audit event.
- Tell the person. The response says what was erased, what was kept, under which law and until when.
Rule 8(3) needs care. Its wording covers the personal data itself as well as traffic data and logs, and its own illustration says a platform must keep order details, personal data and logs for at least one year from the transaction even if the user deletes her account. So an erasure request may not mean immediate deletion of everything: a workable design takes the data out of live use at once, keeps the Rule 8(3) set in the restricted store, and erases it when the year ends unless another law requires longer. How Rule 8(3) sits beside Section 12(3) is a question for counsel. Our post on one-year log retention in the cloud covers the storage side.
For backups, editing snapshots is rarely practical. Keep a hashed list of erased person IDs, reapply erasure on every restore, let backups expire on schedule, and state that schedule in your response.
How do you pass requests on to processors and vendors?
Section 8(7)(b) requires you to cause your Data Processor to erase data you made available to it. Send corrections the same way. Build this into contracts and the router:
- Contract terms. How you send instructions, the time to act, the confirmation sent back, and cover for their sub-processors.
- One sub-task per vendor. Sent by API where they offer one, otherwise by a tracked email to a named address.
- Confirmation as evidence. Store the vendor's reply or API response on the sub-task.
- Escalation. No confirmation by the due date moves the sub-task to an owner, and keeps the request in Partial.
If you are the processor for your own customers, the same flow runs in reverse: you receive their instructions and confirm.
How do nominee requests work under Section 14?
Rule 14(4) lets a Data Principal nominate one or more individuals, in line with your terms of service and applicable law. In practice that means three pieces:
- Nomination capture. An account setting for a nominee's name and contact. The nomination is itself a request record.
- Trigger check. When a nominee asks, verify their identity against the stored details and check evidence of death or incapacity. Section 14(2) defines incapacity as inability to exercise rights due to unsoundness of mind or infirmity of body.
- Same scope. Once accepted, the nominee's requests run through the same router with requester_role set to nominee.
Get counsel to write the evidence list and the wording. Treat this path with care: it is the one place where a mistake hands one person's data to another.
What should the audit log of each request record?
If the Board or a customer asks how you handled a request, answer from the log. Write an append-only event for each change: state transitions, verification results, each sub-task sent and confirmed, each hold created, and each response sent with its template version. Store references, not personal data, in the events.
Which metrics show the workflow is working?
- Volume by type and channel.
- Time to verify and time to close, median and slowest, by type.
- Requests stuck in Partial, and which systems or vendors cause it.
- Grievances near the published period. Treat each as an incident.
- Rejections by reason code. A spike in "identity not verified" suggests the check or message needs work.
- Manual task share. Automate the busiest first.
- Restore reapplications. Every backup restore should show the erasure list reapplied.
Rights-request workflow checklist
- 1. Publish the request channels, the identifier you need, the grievance response period (90 days at most) and the Rule 9 contact.
- 2. Build one request record and route every channel into it.
- 3. Verify against identifiers you already hold; step up only for high-impact actions.
- 4. Build the system register from your data map, with an owner and connector for every system, vendor, backup and AI tool.
- 5. Enforce the state machine; no Done while a sub-task is open.
- 6. Use fixed rejection reason codes and send the reason to the person.
- 7. Implement restrict and retain with reason, with purge dates.
- 8. Reapply erasure on every backup restore.
- 9. Put processor instructions and confirmations in contracts and track them as sub-tasks.
- 10. Add nominee capture and a trigger check.
- 11. Write an append-only audit log and review the metrics monthly.
- 12. Run a drill: one request of each type, answered end to end from records.
The core DPDP duties, including rights, start on 13 May 2027. Start with the register and the request record now. If you want help, DPDP implementation is where we build the router, connectors and audit log in your stack, and a DPDP gap assessment shows which systems are not yet covered.
This article is general information, not legal advice. Our partner law firm gives the written legal opinion on how the Act applies to you.
Sources: Digital Personal Data Protection Act, 2023, Sections 6, 7, 8, 11, 12, 13, 14 and 15 and the Schedule (Gazette of India, 11 August 2023). Digital Personal Data Protection Rules, 2025, Rules 1, 6, 8, 9 and 14 and the Seventh Schedule (G.S.R. 846(E), Gazette of India, 13 November 2025). Last reviewed 19 November 2026.
FAQ
What rights does a Data Principal have under the DPDP Act?
Four rights in Sections 11 to 14: access to a summary of their data and who it was shared with, correction, completion, updating and erasure, grievance redressal, and nominating someone to act for them on death or incapacity. Section 8(7) also requires erasure when consent is withdrawn, unless a law requires retention.
How long do I have to respond to a DPDP rights request?
The Act and Rules set no fixed deadline for access, correction or erasure. Rule 14(3) requires you to publish the period within which you respond to grievances, which must be reasonable and cannot exceed 90 days. Set internal targets for each request type well inside that limit, since unanswered requests become grievances.
Can I refuse to erase data because of KYC or tax rules?
Only for the data a law actually requires you to keep, which your counsel should confirm. Section 12(3) allows retention needed for the specified purpose or to comply with a law. Erase everything else, move the held data to a restricted store, record the legal basis and a purge date, and tell the person what you kept and why. Rule 8(3) also requires personal data, traffic data and processing logs to be retained for at least one year from the date of processing, so ask counsel how that applies to an erasure request.
How should I verify identity for a DPDP request?
Use identifiers you already hold. Rule 14(1)(b) asks you to publish the username or identifier needed to identify the person. Accept requests from a logged-in session, or send a one-time code to the mobile or email on file. Avoid asking for new ID documents, except for nominee requests that need evidence of death or incapacity.
Do DPDP erasure requests apply to my vendors?
Yes. Section 8(7)(b) requires a Data Fiduciary to cause its Data Processor to erase personal data it shared for processing. Put the instruction method, deadline and confirmation in each processor contract, track each vendor as a sub-task of the request, and keep the confirmation as evidence.
When do DPDP rights obligations start?
The rights duties under Sections 11 to 14 and Rule 14 apply from 13 May 2027, eighteen months after the DPDP Rules were notified on 13 November 2025. Some readings count from the day the Gazette was made available, which would give 14 May 2027, so planning for 13 May 2027 is the safe choice.
