Skip to content
EncryptTechBD
Home Industries Our Work Blog Pricing About Contact Discuss Your Project
বাংলা (BN)
Privacy & Compliance 6 min read

Privacy by design checklist for software development

A practical privacy by design checklist covering data mapping, minimisation, consent, retention, access, vendors, testing and evidence.

E
EncryptTechBD · 0 comments
Businessman reviewing a document at a laptop
Photo: Direct Media / StockSnap

Privacy by design is often presented as a legal principle. For a software team it is more practical than that: it is the set of decisions that determine whose data enters the system, why, who can see it, how long it stays and how it leaves. A policy document does not make a system private. A data model, a permission check and a deletion job do.

This checklist is written for teams building or changing a business application in Bangladesh. It assumes personal data is involved somewhere, and that the organisation wants evidence of its decisions rather than a statement of intent. Legal requirements vary by sector and continue to develop in Bangladesh; confirm the current position with counsel before treating this checklist as a compliance statement.

1. Start with a data map

You cannot protect data you have not located. Before writing code, list every personal data element the system will hold and trace it from collection to deletion.

  • Record the data element, the person it belongs to, the purpose, the source and the system of record.
  • Mark the lawful basis or business justification, and whether the data is sensitive.
  • Map every copy: database, file store, search index, analytics tool, backup, log and support export.
  • Name the owner for each element and the retention period.
  • Note every external recipient, including processors and group companies.

Owner: product owner with engineering. Evidence: a data map reviewed at design approval and updated when the schema changes.

2. Purpose, minimisation and defaults

Every field must earn its place. Ask what decision or service the field supports, and what happens if it is left blank. Remove fields collected "for later" or because another system has them.

  • Collect the minimum needed for the stated purpose, at the point it is needed.
  • Use derived values and aggregates where the raw identifier is not required.
  • Set privacy-protective defaults: private profiles, limited visibility, no public sharing by default.
  • Avoid free-text fields for sensitive data unless the process genuinely requires them.
  • Document any decision to retain a field that is not strictly necessary, with the reason and approver.

Owner: product owner. Evidence: field-level justification in the data map and a design review record.

3. Consent and preferences

Consent is not a checkbox at the bottom of a form. It is a record of what a person agreed to, for which purpose and through which channel, with an easy way to change their mind.

  • Separate consent from the terms of service and from other purposes.
  • Record the wording shown, the timestamp, the channel and the version of the notice.
  • Provide an opt-out that works without a support ticket.
  • Propagate preferences to every system that sends messages or shares data.
  • Do not treat silence, a pre-ticked box or a bundled purchase as consent.

Owner: marketing and product. Evidence: consent records, preference centre logs and suppression tests.

4. Retention and deletion

Retention is a design decision, not an archive policy written after launch. For each data element, define how long it is kept, what triggers deletion and what happens to backups.

  • Write the retention period next to the data element in the map.
  • Automate deletion or anonymisation; manual clean-up does not survive a busy quarter.
  • Handle dependent records: a deleted customer may still have invoices that law or accounting requires you to keep.
  • Define backup retention and the process for restoring data that should have been deleted.
  • Test deletion with a real record and confirm it disappears from every copy.

Owner: engineering with the data owner. Evidence: retention schedule, deletion job logs and a completed deletion test.

5. Access and tenant isolation

A privacy incident often begins with an access failure rather than a sophisticated attack. The system must answer "who can see this record, and why?" at the object level, not only at the menu level.

  • Define roles by task and check permissions on the server, not only in the interface.
  • Enforce tenant or branch isolation in every query, export and report.
  • Restrict bulk export and "view all" screens to named roles.
  • Log access to sensitive records and review unusual patterns.
  • Remove access promptly when a person changes role or leaves.

Owner: engineering and IT. Evidence: role matrix, permission tests with a low-privilege account and access review notes.

6. Vendors and data sharing

Every integration is a data-sharing decision. A payment gateway, SMS provider, analytics tool or support desk may receive personal data, and the contract must reflect that.

  • List the data sent to each vendor and the purpose.
  • Confirm the vendor's security and privacy commitments in writing.
  • Prefer the minimum data transfer: send an identifier rather than the full profile where possible.
  • Check where data is stored and whether cross-border transfer is acceptable.
  • Include deletion, breach notification and exit assistance in the agreement.

Owner: procurement with the data owner. Evidence: vendor register, data processing terms and a data-flow diagram.

7. Security controls and privacy testing

Privacy depends on security, but the two are not identical. Encryption and access control protect confidentiality; minimisation and retention limit exposure. Test both.

  • Encrypt data in transit and at rest, and manage keys separately from the application.
  • Hash or tokenise identifiers used in analytics and logs.
  • Keep personal data out of application logs, error trackers and test fixtures.
  • Add privacy cases to the test plan: unauthorised access, export limits, consent withdrawal and deletion.
  • Review third-party scripts and tracking pixels before release.

Owner: engineering and security. Evidence: test cases, configuration review and a release checklist signed by the product owner.

8. Verification and evidence

A checklist is only useful if someone can prove it was applied. At each release, collect the same small set of artefacts:

  • Updated data map and retention schedule.
  • Design review notes showing privacy risks and decisions.
  • Permission test results from a standard user account.
  • Consent and preference test results.
  • Deletion test result and backup handling note.
  • Vendor list with the data shared and the agreement status.

Review the evidence quarterly and after any significant change. If a release adds a new data field, integration or audience, the map and the tests should change with it.

Where this fits in our work

If you need privacy requirements turned into system design rather than a policy appendix, start with Managed Technology and Cybersecurity. The pricing page explains how reviews and implementation are scoped. For a sector example, read how student data privacy obligations change as you grow.

Sources

Services that deliver this

Want this implemented properly?

Pricing →

Every engagement starts with a written scope. If a requirement is outside it, we quote it separately before starting.

Responses (0)

No comments yet. Add the first one below.

Leave a response

Your email address will not be published. Required fields are marked *.

Keep reading

More field notes