UK GDPR. Accountability guide.
ICO accountability framework: what data mapping evidence looks like
Match the ICO’s data mapping, record and access expectations to evidence, then test your record against your systems in both directions.
Co-founder and CEO · Updated November 8, 2026

On this page
Test the map both ways, then file the result
The map passes two tests. A sample of activities from your record each trace to named systems where the personal data was found. A sample of systems holding personal data each trace to an activity. Log any miss with an owner, a decision and a date.
Applies to UK GDPR and the ICO’s Accountability toolkit. Since September 30, 2026 the regulator is the Information Commission. It took over the functions of the Information Commissioner’s Office (ICO) under the Data (Use and Access) Act 2025, GOV.UK reports. The toolkit pages still sit on ico.org.uk, so this guide calls it the ICO’s toolkit.
Roles: the privacy lead owns the record, system owners confirm their systems, and a database administrator runs the queries. Time: allow one working session per sampled activity, plus one for the system-side list.
- Sample activities from your record of processing
- Trace each one to its systems and check the data found
- Tag tables with their activity, then list the untagged ones
- Log every mismatch with an owner and a date
- File the sample, the method and the result together
ICO expectations, evidence and the check it poses
Control measures from the ICO’s Records of processing and Records management and security pages. The third column quotes or closely follows the ICO’s own effectiveness check.
| ICO control measure | Evidence to file | Check the ICO poses | Gap that fails it |
|---|---|---|---|
| Data mapping gives a clear understanding of what information is held and where | A dated map by system: owner, categories, method and date | Would the record match what people were currently doing? | A map built from questionnaires alone |
| The data map is kept up to date, with responsibility assigned | A named owner per area and a log of reviews | Could staff explain their responsibilities in practice? | No owner, so nobody updates the map |
| The record of processing rests on the mapping and is reviewed regularly | Each activity traced to its systems; review dates | Do staff say the process keeps the record correct and current? | Activities listed with no system behind them |
| The record holds every Article 30 item | A reconciliation sheet for each Article 30 item | Would staff say the processes keep the record current and correct? | Erasure limits and security measures left blank |
| The record links to related documents, including the location of personal information | Links from each activity to contracts, DPIAs, breach records and system locations | Is it easy for staff to reach those documents from the record? | Documents held in places the record doesn’t name |
| An asset register lists systems that process personal information | Register with owner, location, retention period and security measures | Given a sample of software, could you show the register’s details are correct? | A register a sample check can’t confirm |
| Access is limited to authorized staff and reviewed regularly | An access export and dated review decisions | Would a sample of new starters, movers and leavers match your procedure? | Access granted years ago and never reviewed |
The two-direction test, filled in
Results for one estate. Rows 1 and 2 start from the record; rows 3 to 6 start from the systems.
| Source | Status | Reason | Owner |
|---|---|---|---|
| Billing database (PostgreSQL) | Matches the record | Record to system: the Invoicing activity names it, and the categories found match the entry | Finance systems owner |
| CRM (SaaS) | Entry out of date | Record to system: entry lists name and email; phone and date of birth also found | Sales operations lead |
| Analytics warehouse | Not in the record | System to record: holds customer extracts, and no activity names it | Data platform lead |
| Test database | Not in the record | System to record: customer rows found, and no activity covers test use | Engineering manager |
| Marketing platform (SaaS) | Not measured | Record to system: the activity names it, but no connection or export exists yet | Marketing lead, by November 30, 2026 |
| Paper archive | Excluded | Outside the electronic checks; a separate review covers it, approved October 2026 | Facilities manager |
Five steps for the system-side list
Step 3 changes catalog metadata and carries its rollback; step 4 is read-only. Both target PostgreSQL 16 to 18.
1. Sample activities from your record
- Owner
- Privacy lead
- Tool or path
- Your record of processing
- Expected output
- A list of sampled activities, each with its systems, including the one with the most sensitive data.
- How to verify
- The sample and the reason for choosing it are written down for the reviewer.
2. Trace each activity and check the systems
- Owner
- Privacy lead, with system owners
- Tool or path
- Catalog queries, scans or a sample of values
- Expected output
- Per activity: systems, categories found, check date and method.
- How to verify
- The categories found equal the categories in the entry, or the difference is logged.
3. Tag each table with its activity
- Owner
- Database administrator, as the table owner
- Tool or path
- psql, as the owner of the table
- Command
- Changes configurationSQLTested on PostgreSQL 16.15, 17.11 and 18.6, November 9, 2026, as the postgres superuser, against synthetic data.
Rollback. Raise a change record before you run the change.-- Raise a change record first. Read the current comment before you overwrite it: -- SELECT obj_description('billing.payment_card'::regclass, 'pg_class'); COMMENT ON TABLE billing.payment_card IS 'ROPA: ACT-04 Invoicing'; -- Rollback: restore the comment you read, or clear it with -- COMMENT ON TABLE billing.payment_card IS NULL;COMMENT ON TABLE billing.payment_card IS NULL; (or the comment you read first) - Expected output
- The table carries a comment that starts with ROPA: and its activity ID.
- How to verify
- Read the comment back with obj_description. It shows the ID you wrote.
4. List the tables that carry no activity tag
- Owner
- Database administrator
- Tool or path
- psql, as a role with catalog access
- Command
- SQLTested on PostgreSQL 16.15, 17.11 and 18.6, November 9, 2026, as the postgres superuser, against synthetic data.
-- Read-only. Lists ordinary and partitioned tables with no 'ROPA: ACT-nn' comment. SELECT n.nspname AS schema_name, c.relname AS table_name, pg_get_userbyid(c.relowner) AS owner_role, obj_description(c.oid, 'pg_class') AS table_comment FROM pg_class AS c JOIN pg_namespace AS n ON n.oid = c.relnamespace WHERE c.relkind IN ('r', 'p') AND n.nspname NOT IN ('pg_catalog', 'information_schema') AND n.nspname !~ '^pg_toast' AND coalesce(obj_description(c.oid, 'pg_class'), '') !~ '^ROPA: ACT-[0-9]+' ORDER BY 1, 2; - Expected output
- One row per table with no activity tag, with its owner role.
- How to verify
- The table you tagged in step 3 is absent. Count the tables a second way, and the two counts agree.
5. Log each miss and file the result
- Owner
- Privacy lead
- Tool or path
- Your reconciliation sheet
- Expected output
- A row per untagged table or mismatched entry: owner, decision (add, delete or justify) and date.
- How to verify
- No row lacks an owner or a date, and the sheet states which systems were not checked.
What the toolkit is and what it isn’t
The toolkit is the ICO’s account of what it looks at when it audits an organization. The legal duties sit in UK GDPR Article 5(2) on demonstrating compliance, Article 24 of the UK GDPR on technical and organizational measures and Article 30 of the UK GDPR on the record. Following the toolkit doesn’t make your processing lawful.
Each toolkit page carries a notice that its guidance is under review after the Data (Use and Access) Act 2025 came into law on June 19, 2025. Check the ICO’s plans for new and updated guidance before you rely on a control measure, and note the date you read it.
What this method misses
The tag query sees only the database you run it in, and only tables. Files, mailboxes, SaaS sources, views and copies in other databases need their own list. A tag is a note that someone wrote, so a wrong tag passes the query. The reviewer checks a sample of tagged tables against the data.
The guide leaves out lawful basis, transparency, impact assessments and transfer mechanisms, though each links to the record. It also leaves out other changes in the 2025 Act, which come into force in stages. This page is general information, not legal advice.
IRIS gives the system side of the test, in the sources you connect. The record stays with your privacy team. IRIS produces evidence that supports your assessment. Capabilities vary by edition.
A system-side inventory of sensitive data, with each finding’s source and scan date: IRIS produces part of this, in the sources you connect.
Who can reach that data: IRIS produces part of this, where the source exposes permission data.
The reason for each table or file a scan couldn’t read: IRIS produces part of this, for the tables and files the scan account can list.
Your record of processing activities: Not produced by IRIS.
A scored readiness view for UK GDPR: not in v1; ask us if you need one: Not produced by IRIS.
Explore audit evidenceRun the test on one activity this month
Choose the activity that holds the most sensitive data and trace it to its systems. The test passes when the categories found equal the entry, the untagged-table list holds nothing unexplained and every mismatch has an owner and a date.
For the item-by-item comparison that follows, use GDPR Article 30 records versus the real data estate.
Official sources
- ICO Accountability toolkit: records of processing and lawful basis, Information Commissioner’s Office. Opened November 8, 2026.
- ICO Accountability toolkit: records management and security, Information Commissioner’s Office. Opened November 8, 2026.
- GOV.UK news story: Information Commission succeeds the ICO as UK’s data protection regulator, September 30, 2026, GOV.UK. Opened November 8, 2026.
- Data (Use and Access) Act 2025, legislation.gov.uk. Opened November 8, 2026.
- UK GDPR Article 30: records of processing activities, legislation.gov.uk. Opened November 8, 2026.
- PostgreSQL 18: COMMENT and the object information functions, PostgreSQL Global Development Group. Opened November 8, 2026.
Start with a free assessment of one database you choose.
Get a free assessmentICO accountability questions
No. It sets out what the ICO looks at in audits. The legal duties sit in UK GDPR, including Article 5(2) and Article 30.
The Information Commission, since September 30, 2026. It took over the Information Commissioner’s Office’s functions under the Data (Use and Access) Act 2025, and the toolkit pages remain on ico.org.uk.
Yes, in the ICO’s view. Its toolkit expects the record to be based on a data mapping exercise and reviewed regularly.
Yes, in part. The toolkit lists software that maps information flows through IT systems automatically as an option to consider. People still own purposes, lawful basis and the review.
Protect your brand. Start with your data.
A free assessment scans one database, read-only, and shows what it holds, who can reach it and what it couldn’t check.


