Records of processing that hold up: building from systems, not spreadsheets
The usual record of processing is a spreadsheet filled in once, by someone who was asked to, from memory. It is out of date by the next quarter. A better one starts at the systems.
Key points
- A record of processing activities is a written inventory of what personal data an organization handles, why, who receives it, where it goes, how long it is kept and how it is protected, one entry for each processing activity.
- A record built from systems starts at the databases, applications and vendors that actually hold the data and works back to the business activity, instead of asking each department to describe its work from memory.
- A usable record covers, for each activity, the purposes, the categories of people and of personal data, the recipients, transfers to other countries, retention periods and the security measures in place.
- Records go stale because nothing connects them to change: a new vendor, a new field or a new integration is added and nobody updates the sheet. A record stays accurate when changes to systems trigger a change to the record.
What the law asks the record to contain
Article 30 of the GDPR, on EUR-Lex, requires controllers to keep a record of their processing activities. The UK regulator’s documentation guidance sets out what that record covers: the organization’s name and contact details, the purposes of the processing, the categories of individuals and of personal data, the categories of recipients, details of transfers to other countries with the safeguards used, retention schedules, and a description of the technical and organizational security measures. It adds that the record must be in writing, granular, meaningful and kept up to date.
Whether your organization must keep the record, and whether any exemption applies, is a legal question for your counsel. Other regimes ask for similar inventories under other names, and a good record serves them too: it is the document from which notices, requests, impact assessments and vendor reviews are built.
Why spreadsheets drift
Records go stale because nothing connects them to change: a new vendor, a new field or a new integration is added and nobody updates the sheet. A record stays accurate when changes to systems trigger a change to the record.
Three habits produce most of the drift.
- One row per department, not per activity. “HR” is not an activity. Recruitment, payroll and performance review have different purposes, data, recipients and retention, so they need different rows.
- Purposes written as slogans. “Business operations” cannot be tested. A purpose should be specific enough that you can tell when a new use falls outside it.
- No link to the system. If the entry does not name the application, database and vendor, nobody can tell when a system change makes it wrong.
Build it from the systems
A record built from systems starts at the databases, applications and vendors that actually hold the data and works back to the business activity, instead of asking each department to describe its work from memory. The inventory of systems is shorter and more stable than the list of things people do with them, and it is the list that a vendor invoice, a single sign-on log or a cloud account will confirm. The ICO’s guidance recommends questionnaires, meetings with the business and a review of policies and contracts to find out what is held; the systems approach adds the technical side to those methods.
The sources, in the order we usually go to them:
- The single sign-on or identity provider’s application list.
- Cloud accounts and their data stores.
- The accounts-payable ledger, to find vendors nobody mentioned.
- Contracts and data processing agreements.
- Interviews with each system’s business owner, to turn each system into one or more activities.
Finding the data itself, including the places no inventory lists, is a separate task covered in where personal data hides.
The fields, and where each one comes from
A usable record covers, for each activity, the purposes, the categories of people and of personal data, the recipients, transfers to other countries, retention periods and the security measures in place. Each of those fields has a natural source, which is what makes the record maintainable.
This is why a record from systems is cheaper to keep current: five of the six fields can be checked against something real. The retention field in particular should point at a retention schedule, not repeat a number that will drift.
Ownership and change
Each entry needs a named owner in the business, not the privacy lead, because the owner is the person who knows when the activity changes. The owner confirms the entry at an agreed interval and when the activity changes. The privacy lead reviews, but does not write.
The cheapest control is a trigger. If a new vendor cannot be paid until it has an entry, and a new integration cannot be released until its entry is checked, the record stays accurate without a yearly clean-up. A privacy impact assessment for a new project updates the record as its last step. See privacy impact assessments.
A short test of your record
- Pick a vendor from the ledger at random. Is it in the record?
- Pick an entry at random. Can the owner say when it last changed?
- Take an access request from last quarter. Could the record have told you where to search?
- Take the newest integration. Did the record change before it launched?
If three of the four fail, the record is a document, not a control. The method is described on data mapping and records of processing.
How the work is bounded
The scope is agreed in writing before work starts, and the engagement is quoted in writing with it.
SecHB does not issue certifications, attestations or audit opinions: those come from accredited certification bodies, CPA firms and QSAs. The work here is what an organization does to be ready for them.
Nothing here is legal advice. Where a question turns on the law, the work is done alongside the client’s counsel, not instead of them.
Questions we are asked
What is a record of processing activities?
A record of processing activities is a written inventory of what personal data an organization handles, why, who receives it, where it goes, how long it is kept and how it is protected, one entry for each processing activity.
What does building a record from systems mean?
A record built from systems starts at the databases, applications and vendors that actually hold the data and works back to the business activity, instead of asking each department to describe its work from memory.
What fields should each entry contain?
A usable record covers, for each activity, the purposes, the categories of people and of personal data, the recipients, transfers to other countries, retention periods and the security measures in place.
Why do records go stale?
Records go stale because nothing connects them to change: a new vendor, a new field or a new integration is added and nobody updates the sheet. A record stays accurate when changes to systems trigger a change to the record.
Who should own each entry?
Each entry needs a named owner in the business, not the privacy lead, because the owner is the person who knows when the activity changes.
Build the record once
Tell us which systems hold personal data and where the record lives today. The reply says what a system-first build would involve. The service is data mapping and records of processing; the first quarter of a privacy lead is in what a fractional DPO does in the first 90 days.