Control call recordings, identity checks, KYC discussion, screen access, agent scripts, remote work, quality review, retention and complaint evidence.
A call centre can expose more personal data in one conversation than the customer ever entered in an app.
Limit what agents can see, say, record and export, and build scripts that never request secrets.
Map call types and data fields.
Approved scripts.
Full account data visible to every agent.
The Digital Personal Data Protection Act, 2023 and the final Rules notified in November 2025 follow phased commencement. As of 25 June 2026, organisations should separate duties already operative from consent, grievance, rights, children, Significant Data Fiduciary and other operational provisions scheduled for later commencement, while continuing to comply with the IT Act, CERT-In directions and sector-specific rules already in force.
Notice, purpose, security and grievance requirements should be built into call recording and customer-support operations as DPDP provisions commence.
Current banking and fraud-protection practices already require customers not to disclose passwords, PINs or OTPs, and support teams must reinforce that boundary.
| Area | What to establish | Operating rule |
|---|---|---|
| Recording | Purpose, notice and retention. | Avoid perpetual storage. |
| Verification | Risk-based questions and account signals. | Never request secrets. |
| Access | Agent screen, masking and download. | Use least privilege. |
| Quality | Script testing and complaint review. | Monitor misuse. |
A reliable control should connect the individual, data field, purpose, notice or sector disclosure, system, employee access, vendor access, retention rule and closure evidence. A policy statement that cannot be traced through this chain is difficult to operate.
Maintain a legal-timing matrix. Record the DPDP provision, phased commencement status, current IT Act or sectoral duty, business owner, system dependency and implementation deadline. Avoid one blanket label such as compliant or not compliant.
Build controls into technology and workflow. A written instruction cannot stop an SDK from collecting contacts, a campaign tool from re-importing suppressed users or an agent from downloading medical records unless the system enforces the decision.
Use proportionate verification. Weak checks can expose another person’s information; excessive checks create more Aadhaar, health, payroll or bank data that must be protected and deleted later.
Generate evidence during ordinary operations: versioned screens, event logs, access approvals, vendor tickets, complaint chronology, deletion reports, test recordings and management decisions.
Run a negative-path test: refusal, withdrawal, account closure, vendor breach, employee exit or child-user flow. The control should continue to protect data outside the happy path.
Management reporting should show overdue actions, repeat complaints, failed tests and residual risk rather than only the publication of policies.
Select one real user or transaction journey and trace it from collection through sharing, access, retention, withdrawal, complaint or closure. Capture the evidence at each stage.
Test the control on production-like systems rather than screenshots alone. Review network traffic, event logs, suppression status, vendor responses, role access and deletion output.
Run an adverse scenario: the vendor is breached, the user is a child, the borrower alleges harassment, the employee leaves or the app permission is revoked. Record the response and gaps.
Compare public wording with actual behaviour. Product forms, call scripts, privacy notices, contracts, SDKs and support tools should tell the same story.
Assign a named owner, funded action and closure date to each gap. Retain the reason when management accepts residual risk or chooses a less intrusive alternative.
Start with the privacy, security, product or regulated-business owner and preserve system evidence before changing configuration or deleting records. Separate current sector and CERT-In obligations from future DPDP readiness.
For serious complaints, children’s data, financial harassment, medical exposure or suspected cybercrime, involve qualified legal, privacy, cyber, banking, insurance or healthcare specialists and use the applicable official channel.