Quick answer
A ROPA (record of processing activities) is the written inventory GDPR Article 30 asks controllers and processors to keep: why personal data is processed, whose data and which, who receives it, where it goes, how long it is kept and how it is secured. The two roles keep different records.
What is a ROPA?
A ROPA, also written RoPA, is a record of processing activities: the written inventory that Article 30 of the GDPR asks controllers and processors to keep, covering why personal data is processed, whose data and which data, who receives it, where it goes, how long it is kept and how it is protected. (If you were looking for the Spanish word ropa, clothing, this is not that page.)
Three rules from the text of Article 30 shape everything else. The record must be in writing, and electronic form counts (Article 30(3)). It must be made available to the supervisory authority on request (Article 30(4)). And it exists, in the words of Recital 82, so that a controller or processor can show compliance by keeping track of its processing. The idea behind it is practical: a team that cannot fill in the rows does not yet know what it does with personal data. This page explains the concept; it is not legal advice.
Who has to keep a ROPA?
Both controllers and processors. Article 30(5) exempts an enterprise or organisation with fewer than 250 employees from the duty, but the exemption falls away if any of the following is true:
| Condition | In plain words |
|---|---|
| The processing is likely to result in a risk to the rights and freedoms of people | A real risk, not a theoretical one |
| The processing is not occasional | It happens regularly or continuously |
| The processing includes special categories of data (Article 9(1)) or data on criminal convictions and offences (Article 10) | Health, beliefs, biometrics and similar; criminal records |
Recital 13 says the exemption exists to take account of micro, small and medium-sized enterprises. A tracking script runs on every visit, so "not occasional" is hard to rule out for a live website; the French regulator, the CNIL, recommends that small organisations include borderline processing in their record anyway. Whether you are exempt is a question for your data protection officer or counsel.
What goes in a controller's record
Article 30(1) lists seven items. The table shows each one as a question, with a filled-in example for a website analytics activity (sample values):
| Item | Question | Example for website analytics |
|---|---|---|
| (a) Contact details of the controller, any joint controller, representative and DPO | Who is responsible? | Example Ltd, privacy@example.com |
| (b) Purposes | Why is the data processed? | Measure which pages and campaigns bring visitors; find and fix errors |
| (c) Categories of data subjects and of personal data | Whose data, and which? | Website visitors; pseudonymous visitor ID, page URL, referrer, country, device type |
| (d) Categories of recipients, including in third countries | Who receives it? | Analytics provider acting as processor; the marketing team |
| (e) Transfers to a third country, with safeguards where required | Where does it go? | None, or the named country and the safeguard used |
| (f) Time limits for erasure, where possible | How long is it kept? | Events 12 months; user profiles 24 months |
| (g) General description of security measures, where possible | How is it protected? | Address truncation, personal-data stripping, role-based access |
Article 30 does not list the lawful basis for processing, although many templates add a column for it. You still have to decide the basis for each activity; Article 6(1) names six, such as consent and legitimate interests.
Controller record vs processor record
A processor keeps a shorter record under Article 30(2), organised by controller:
| Item | Controller, Article 30(1) | Processor, Article 30(2) |
|---|---|---|
| Contact details | Controller, joint controller, representative, DPO | Processor, each controller it acts for, representatives, DPO |
| Purposes | Yes | Not listed; instead, the categories of processing carried out for each controller |
| Data subjects and data categories | Yes | Not listed |
| Recipients | Yes | Not listed |
| Transfers to third countries | Yes | Yes |
| Erasure time limits | Where possible | Not listed |
| Security measures | Where possible | Where possible |
One organisation can be both: a software company is a controller for its own staff and customer-account data and a processor for the data its customers put into the product. The CNIL recommends keeping two separate records in that case.
Worked example: a ROPA row and two checks
Sample data: a small online shop keeps a ROPA with 5 activities (orders, customer support, newsletter, website analytics, staff payroll). Two quick checks show whether it is usable.
Check 1, completeness. Each activity needs the 7 controller items. 5 × 7 = 35 fields. A review finds 9 blank: 4 retention periods, 3 security descriptions and 2 transfer entries. Completeness = (35 − 9) ÷ 35 = 74.3%. Each blank is a question nobody has answered yet.
Check 2, the orphan-property test. Take the website analytics activity and list every event property the site sends: 40. Map each one to the purpose written in the row. 36 map to "measure which pages and campaigns bring visitors"; 4 do not (a free-text comment field, a birth date and two others). Coverage = 36 ÷ 40 = 90%, with 4 orphans. An orphan is data collected for no recorded purpose: either stop collecting it, which is what data minimization asks for (Article 5(1)(c)), or add a purpose and a basis for it.
| Check | Result | What to do |
|---|---|---|
| Completeness | 26 of 35 fields filled (74.3%) | Assign an owner to each of the 9 blanks |
| Property coverage | 36 of 40 mapped (90%) | Remove or justify the 4 orphans |
ROPA vs data map vs privacy policy
| Document | Written for | Level of detail | Typical trigger to update |
|---|---|---|---|
| ROPA | The regulator on request, and your own team | One row per processing activity | A new purpose, tool, recipient or retention period |
| Data map or inventory | Your own team | Systems, fields and flows; see data lineage | A new system or integration |
| Privacy policy | The people whose data it is | Plain language, public | A change in what you tell people |
A data map is usually the input and the ROPA the output; both belong to wider data governance. The three should never contradict one another.
Common ROPA mistakes
- One giant row. "Marketing" as a single activity cannot answer purposes, recipients and retention separately. Split by purpose.
- Copying retention from a policy, not from the system. Record the period that is actually configured.
- No owner and no review date. The record goes stale the first time a tool is added without anyone updating it.
- A one-off document for an audit. Article 30 asks for a record you keep, not a file you produce once.
- Leaving "where possible" fields blank with no note. Items (f) and (g) have the qualifier, so if you cannot fill them, write why.
- Assuming the exemption. Check the three conditions above before you decide you do not need one.
ROPA and TapCub
TapCub does not generate or store a ROPA. For the website analytics row of the example, though, several settings answer fields directly, so you can copy facts instead of guessing.
- Purposes and data categories. In Configuration › Privacy center, consent has three purposes (analytics, functional, marketing), and the collection tier tells you what is collected: tier 1 anonymous aggregate with no persistent ID, tier 2 cookie-free pseudonym, tier 3 identified with a device ID and login. The compliance center is where those are set.
- Every property, classified. In All-platform analytics › Data › Property dictionary, each property is classed as a direct identifier, an indirect identifier or non-personal, and Tracking plan lists planned events and their properties. These two lists are the input for the orphan-property test.
- Time limits (f). The defaults in the compliance configuration are 12 months for events, 90 days for autocapture and 24 months for profiles and consent receipts, capped by your plan; a scheduled job deletes expired rows.
- Security measures (g). The address is not written to the event: it is used for location and the visitor hash, then discarded. Address truncation, personal-data stripping and member-level masking are configurable.
- Evidence of change. Configuration › Settings › Audit log records privacy configuration changes and exports as CSV, NDJSON or JSON, up to 50,000 rows at a time.
- What TapCub cannot fill. Contact details (a), recipients (d) and transfers (e) depend on your contracts and where you chose to host data. Write those from your own agreements.
For the wider picture across laws, see GDPR, CCPA and PIPL.
Sources
- Regulation (EU) 2016/679 (GDPR): Article 30 (records of processing activities), Article 5(1)(c), Article 6(1), Recitals 13 and 82.
- CNIL, Record of processing activities: who keeps a record, small organisations, separate controller and processor records.
Was this explanation clear?
We use your answers to fix and expand this page.
Couldn’t send. Please try again in a minute.
Thanks for the feedback. We’ll use it to improve this page.You’ve already sent feedback on this page. Thanks.

Data Engineer
Writes about event design, data quality, and collection workflows, with verifiable implementation methods for reliable analytics.