Home · About · Services · Industries · Projects · Products · Process · Blog · Glossary · FAQ · Contact
DPDP Act 2023 for software teams: a practical checklist A product-level reading of India’s DPDP Act 2023: consent, notice, purpose, retention, breach response and user rights mapped to concrete changes.
A product manager at a Pune ed-tech company gets an email from the legal team. It says the app must be "DPDP compliant" and asks engineering for a plan by month end. The engineers open the Act, read forty-odd sections of legal text, and still cannot say which tables, screens or jobs need to change.
This article is for that team. It is not legal advice, and your counsel should sign off the final interpretation. What follows is how the Digital Personal Data Protection Act, 2023 translates into backlog items. As of mid-2025, the Act is passed but not yet in force, and the government published draft rules for consultation in January 2025. The final rules will settle timelines and some details. The core obligations in the Act itself are clear enough to start building now.
Who the Act calls what The Act uses its own vocabulary, and it helps to map it to your system early.
Data principal: the person the data is about. Your users, students, patients or customers. Data fiduciary: the organisation that decides why and how data is processed. Usually you. Data processor: anyone processing data on your behalf. Your cloud host, email provider, SMS gateway, analytics tool and outsourced support desk. Significant data fiduciary: a fiduciary the government notifies based on volume, sensitivity and risk. These carry extra duties such as a Data Protection Officer, audits and impact assessments. Consent manager: a registered entity through which principals can give, manage and withdraw consent. The Act covers digital personal data, and non-digital data that is later digitised. It also applies to processing outside India when it relates to offering goods or services to people in India. Unlike some other regimes, it does not define a separate class of sensitive personal data. Children’s data does get special treatment, covered below.
Consent and notice: the screens that change Processing needs either consent or one of the "legitimate uses" the Act lists. These include data a person voluntarily provides for a specified purpose, legal obligations and certain employment purposes. For most consumer products, consent will be the main basis.
Consent under the Act must be free, specific, informed, unconditional and unambiguous, given through a clear affirmative action. In product terms:
No pre-ticked boxes. The user must take an action. Separate consent for separate purposes. Agreeing to receive course updates is not agreement to marketing from partners. A notice before or with the consent request. It must say what data you collect, why, how to withdraw consent and how to complain to the Data Protection Board. The notice must be available in English or any language in the Eighth Schedule of the Constitution, at the user’s option. Plan for translated notice text in your CMS. Withdrawal must be as easy as giving consent. If consent took one tap, withdrawal should not need an email to support. The engineering change is a consent ledger. Store each consent event as a row, not a boolean on the user record.
CREATE TABLE consent_events (
id BIGSERIAL PRIMARY KEY,
user_id BIGINT NOT NULL REFERENCES users(id),
purpose_code TEXT NOT NULL, -- e.g. 'course_updates', 'marketing_email'
notice_version TEXT NOT NULL, -- which notice text the user saw
language TEXT NOT NULL, -- 'en', 'hi', 'mr', ...
action TEXT NOT NULL CHECK (action IN ('granted', 'withdrawn')),
channel TEXT NOT NULL, -- 'web', 'android', 'ios', 'consent_manager'
created_at TIMESTAMPTZ NOT NULL DEFAULT now()
);
CREATE INDEX ON consent_events (user_id, purpose_code, created_at DESC);The current state for a purpose is the latest event. Every service that sends marketing, shares data or runs analytics should check this table, or a cached view of it, before acting.
Purpose limitation and data minimisation Consent is tied to a stated purpose. Using the same data for a new purpose needs a fresh basis. This catches many teams out when product ideas evolve faster than notices.
Practical steps:
Build a data inventory. List each personal data field, where it is stored, which purpose it serves and which processors receive it. Tag fields with purpose codes in your schema documentation. A column with no purpose is a column to drop. Review analytics and third-party SDKs. Many mobile apps ship SDKs that collect device identifiers for purposes nobody wrote down. Add a purpose check to the design review template. Any feature that uses existing personal data in a new way gets flagged. Retention and erasure The Act requires fiduciaries to erase personal data once the purpose is no longer served, or consent is withdrawn, unless the law requires retention. The draft rules propose specific retention periods for some classes of large platforms. For everyone else, you need your own defensible policy.
Retention is where engineering effort concentrates, because data hides in many places.
Primary databases, including soft-deleted rows. A deleted_at flag is not erasure. Read replicas and reporting databases, which inherit deletes only if replication covers them. Object storage: uploaded documents, ID scans, profile photos. Logs. Application logs often contain emails, phone numbers and request bodies. Backups. Most teams accept that backups expire on their own schedule, and document that restored backups re-apply pending deletions. Processors. Your contracts should require processors to delete data when you instruct. Write a scheduled job that enforces retention rules per purpose, and log what it deletes in counts, not contents. Mask or drop personal data in logs at source, rather than trying to clean log stores later.
Breach response Under the Act, a fiduciary must notify both the Data Protection Board and each affected data principal of a personal data breach. The draft rules propose an initial intimation without delay and a detailed report within 72 hours. Treat those as working assumptions until the rules are final.
Separately, CERT-In’s 2022 directions already require reporting specified cyber incidents within six hours. Your incident runbook needs both paths.
What engineering must provide:
Audit logs that answer "who accessed what, and when" for personal data. Without these, you cannot scope a breach. A way to list affected users quickly from logs and database queries. Notification templates in the languages you support, and a tested way to send them in bulk. Contact details and on-call roles agreed in advance with legal and leadership. The Act also requires reasonable security safeguards. Encryption at rest and in transit, access control by role, MFA for admin consoles and secrets kept out of code are the baseline. Penalties for failing to take reasonable safeguards can reach ₹250 crore, so this is not a box-ticking item.
Data principal rights and grievances Principals get rights that need product features, not just policy text.
Access: a summary of personal data processed and the processing activities, plus the identities of other fiduciaries and processors it was shared with. Correction, completion, updating and erasure of their data. Grievance redressal: a way to raise complaints, with a response within a set period. The draft rules propose an outer limit, which you should confirm once final. Nomination: naming someone to exercise rights on their behalf in the event of death or incapacity. A self-service privacy page in the account area covers most of this. Back it with an internal admin queue for requests that need human review, with timestamps so you can prove response times.
Children’s data For users under 18, the Act requires verifiable consent of a parent or lawful guardian. It also prohibits tracking, behavioural monitoring and targeted advertising directed at children, subject to exemptions the government may notify.
For ed-tech, school and coaching platforms this is the largest change. You need an age gate at signup, a parental consent flow and a way to verify the parent. The draft rules suggest verification can rely on reliable identity details already held or on government-issued virtual tokens. Also review analytics and ad SDKs inside any app children use. Many of them do exactly the tracking the Act restricts.
A checklist to take into sprint planning Assign an owner for the data inventory and finish it before writing code. Replace consent booleans with a consent ledger keyed by purpose and notice version. Rewrite notices in plain language and plan translations for your user base. Make withdrawal a one-step action in the account area. Implement retention jobs per purpose and remove personal data from logs at source. Add audit logging on personal data access and test that you can scope a breach in hours. Build the privacy page for access, correction, erasure, grievances and nomination. Add age gating and parental consent if minors can sign up. Review processor contracts for deletion and breach notification clauses. Revisit the plan when the final rules are notified, and adjust timelines and periods. None of these items is exotic. They are ordinary product and backend work. Starting before the rules are final costs little, and it avoids a rushed rebuild when the compliance date arrives.
Infigini IT Services · GSTIN 05AAIFI1118B1Z3 · 11 Shree Ram Puram Lane No.2, Kanwali Road, Dehradun, Uttarakhand 248001, India · info@infigini.com · +91 95684 87789 · LinkedIn