A car-sharing provider stores the start, destination and GPS trace of every trip, linked to the customer account. When someone cancels their contract, the account is deactivated. The trips remain in the analytics database, year after year, although fleet planning only wants to know how many trips start in which district at what time. To this day, the movement profiles would reveal where former customers live and work. A.1.4.6 requires personal data to be deleted, or de-identified so that nobody can trace it back to the person, as soon as the original purpose no longer needs it.
The control marks the end of the life cycle. It asks when processing is over and what then happens to the data.
What does the standard require?
The control sets a point in time and allows two routes.
- The point in time. What counts is the moment when the original data is no longer needed for the defined purposes. From then on the obligation applies, without any further prompt from outside.
- Route 1: delete. The organization has mechanisms for removing data once no further processing is planned. Mechanisms here means established procedures or functions that work reliably, even when nobody thinks of it in the individual case.
- Route 2: de-identify. As an alternative, de-identification techniques may be used. The bar is high: the result must no longer reasonably allow the individuals to be re-identified. Anyone who cannot clear that bar with confidence has to delete.
In practice
Define the triggers. For each processing activity, determine when the purpose has been fulfilled and how much time may pass afterwards before deletion. Typical triggers are cancellation, a closed application process, withdrawal of consent or the expiry of a retention period.
Implement deletion technically. Automated deletion jobs in the leading systems have proven themselves, supplemented by a list of all places where copies arise: exports, file shares, analytics data sets, email mailboxes. Whatever cannot be automated becomes a fixed task with a responsible person and a due date.
Test anonymization before relying on it. Aggregating into groups, removing identifiers, coarsening place and time: every technique has limits. Test the result against other available data; with location data, a handful of points is often enough to recognize a person. Document the method, the test and the known limits.
Include backups. Define how long backups are kept and how deleted data is handled within them. It is common to let it expire with the rotation, combined with the rule that deletions made in the meantime are repeated after every restore.
Prove deletion. Record which data sets were deleted or anonymized when, and by whom. A Deletion Evidence Log with date, system, data category and method is enough for most audits.
Typical audit evidence
For A.1.4.6, auditors typically expect:
- Deletion concept with triggers and periods per processing activity (→ Data Deletion, Masking & Leakage Prevention Policy)
- Configuration and logs of automated deletion jobs (→ Deletion Evidence Log)
- Description of anonymization methods with an assessment of the re-identification risk
- Backup concept with backup retention periods and rules for restores (→ A.8.13 Information Backup)
- Samples showing that data from completed processing activities has actually gone
KPI
% of completed processing activities with timely de-identification or deletion
A processing activity is completed once its trigger has occurred, for instance when the contract has ended. Timely means: within the time the deletion concept allows. The target is 100%. It makes sense to measure with samples in the systems, since the number of scheduled deletion jobs alone says little.
Supporting KPIs:
- Share of systems with an automated deletion job
- Number of data sets past their deletion deadline
- Share of anonymized data sets with a documented re-identification risk assessment
Link to the GDPR
A.1.4.6 puts storage limitation under Article 5(1)(e) GDPR into practice: data in a form that permits identification may only be kept as long as the purpose requires. Article 17(1)(a) obliges the controller to erase data without undue delay once it is no longer necessary for its purpose, and this applies even without a request from the data subject. Article 17(3)(b) covers the exception for legal retention obligations. That effectively anonymized data no longer falls under the Regulation follows from Recital 26 (→ GDPR General Data Protection Regulation).
Annex D of the standard maps A.1.4.6 to three further provisions on pseudonymization and identifiability: Article 6(4)(e) names encryption and pseudonymization as safeguards when assessing further processing, Article 32(1)(a) names both as security measures, and under Article 11(1) the organization does not have to keep additional information only to be able to identify a person for the purposes of the Regulation.
Related controls
A.1.4.6 closes the life cycle, and the neighbouring controls provide periods and methods:
- A.1.4.5 PII minimization objectives: sets which de-identification techniques are used.
- A.1.4.8 Retention: defines the periods after which processing ends.
- A.1.4.9 Disposal: governs how media and documents are destroyed securely.
- A.1.3.7 Access, correction or erasure: handles erasure requests from data subjects.
- A.8.10 Information Deletion: provides the technical requirements for deletion in the ISMS.
Sources
- ISO/IEC 27701:2025, Annex A, Table A.1, A.1.4.6, and Annex B, B.1.4.6: PII de-identification and deletion at the end of processing
- Regulation (EU) 2016/679 (GDPR), Articles 5, 6, 11, 17 and 32 and Recital 26
- ISO/IEC 27701: Privacy Information Management System: structure of the standard and roles