A retail chain plans to roll out video analytics in its stores. The software counts visitors, measures how long they linger in front of shelves and recognizes returning faces. The contract is signed, the cameras are mounted, go-live is set for next week. Only then does someone ask whether a privacy impact assessment exists. Large-scale systematic monitoring of publicly accessible areas plus biometric data: the check should have happened before the first purchase order. A.1.2.6 requires the organization to settle, before any new or changed processing, whether a privacy impact assessment is needed and to carry one out where it is.
The impact assessment is the early warning system of privacy. It only works while the processing can still be shaped. After go-live, risks can usually only be described.
What does the standard require?
The control has two steps: check the need, then carry out the assessment where it is appropriate.
- Check with every change. Whenever new processing or a change to existing processing is planned, the organization checks whether an impact assessment is called for.
- Assess risks to data subjects. The standard assumes that every processing activity creates risks for the people concerned. Some jurisdictions specify the cases in which an impact assessment is mandatory.
- Example triggers. It names automated decisions with legal effects on individuals, large-scale processing of special categories such as health, genetic or biometric data, and large-scale systematic monitoring of publicly accessible areas.
- Define the components. The organization decides what a complete impact assessment has to contain. Typical items are the types of data processed, where they are stored and where they may be sent. Diagrams of the data flows and maps of the data help, and the records under A.1.2.9 supply much of the information.
- Methodical support. For the approach, the standard refers to ISO/IEC 29134.
In practice
Build the check into existing routines. Screening belongs in project kick-off, in the purchase of new software and in change management. A short questionnaire with the criteria from Article 35(3) GDPR and the supervisory authority’s list is enough for the first decision.
Use the authorities’ criteria. In Germany, the Data Protection Conference has published a list of processing operations that always require an impact assessment. The WP 248 guidelines, endorsed by the European Data Protection Board, name nine criteria. If two of them are met, an impact assessment is usually indicated.
Use a fixed template. Define the components once as a template: description and purposes, lawful basis, data types, storage locations, transfers, recipients, deletion periods, necessity, risks to the individuals with likelihood and severity, measures, residual risk, opinion of the data protection officer and approval. Two people assessing two activities then reach comparable results.
Draw the data flow. A diagram from collection to deletion reveals gaps that stay hidden in prose. A route that nobody can draw is usually one that nobody fully understands.
Track measures and repeat. The agreed measures move into risk treatment with a deadline and a responsible person. If the processing changes, the assessment is reviewed. If a high risk remains despite the measures, the supervisory authority has to be consulted before the start.
Typical audit evidence
For A.1.2.6, auditors typically expect:
- Screenings for new and changed processing, including those with a negative result
- Impact assessment reports following a fixed template, with the opinion of the data protection officer and approval
- Data flow diagrams for the processing assessed
- Tracking of measures with deadlines and owners (→ Risk Treatment Plan, Risk Register)
- Anchoring in the project approach (→ Project Management Security Policy)
KPI
% of high-risk processing activities with completed privacy impact assessment
Completed means: report finished, measures agreed, approval given, all before the processing starts. The target is 100%. The figure is only as good as the detection of high-risk cases, which is why the next indicator belongs with it.
Supporting KPIs:
- Share of new projects and software purchases with a documented screening
- Share of agreed measures implemented on time
- Share of impact assessments reviewed after a change to the processing
Link to the GDPR
A.1.2.6 corresponds to Article 35 GDPR. Under paragraph 1, the assessment is carried out before the processing where a high risk is likely. The three examples in the standard largely match the standard cases in Article 35(3)(a) to (c). There, the GDPR also covers data on criminal convictions under Article 10 and, in point (a), lets decisions that similarly significantly affect the person suffice as well. Article 35(4) obliges supervisory authorities to publish lists, paragraph 7 sets the minimum content, paragraph 2 the advice of the data protection officer and paragraph 11 the review after changes. If a high residual risk remains, prior consultation under Article 36 follows (→ GDPR General Data Protection Regulation).
Related controls
The impact assessment pulls together information from many controls:
- A.1.2.9 Records related to processing PII: supplies data types, recipients and measures and holds the report.
- A.1.2.3 Identify lawful basis: settles the basis that the assessment evaluates.
- A.1.3.11 Automated decision making: covers one of the typical triggers.
- A.1.4.2 Limit collection: implements data minimization measures that often result from the assessment.
- A.5.8 Information Security in Project Management: anchors screening in project kick-off.
- A.8.25 Secure Development Life Cycle: carries the results into development and acceptance.
Sources
- ISO/IEC 27701:2025, Annex A, Table A.1, A.1.2.6, and Annex B, B.1.2.6: Privacy impact assessment
- Regulation (EU) 2016/679 (GDPR), Articles 35 and 36
- ISO/IEC 29134: Guidelines for privacy impact assessment
- ISO/IEC 27701: Privacy Information Management System: structure of the standard and roles