Cookies on this site. Help us understand what works and improve your experience. Essential cookies keep the site working. With your permission, our visit insights and Google Analytics measure usage, and Google Ads measures advertising. Optional cookies are not required to use the free kit. Change your choice anytime in Cookie settings. Privacy policyCookie policy
New in 2022: mask, pseudonymise or anonymise data where appropriate, especially personal data.
Is it mandatory? Not automatically. You must consider this control in your Statement of Applicability (6.1.3 d). Include it if your risk assessment, laws, contracts or business needs require it; if you exclude it, write down a justification an auditor will accept.
Why this control matters
Data masking is most effective when teams choose the least revealing method that still supports the business purpose. Owners should distinguish masking, pseudonymisation, and anonymisation, then document where each method is acceptable, who approves exceptions, and how re-identification risk is controlled over time.
Common auditor findings to check for
Potential gaps to test in your own environment, not a claim about observed audit frequency.
Production data is copied into non-production systems, but masking rules do not cover free-text fields, identifiers, or linked tables that still expose personal information.
Support staff can view full records because role-based masking is missing or inconsistently configured across applications, exports, and reports.
Data sets are described as anonymised without testing re-identification risk, sample joins, or indirect identifier combinations that could still reveal individuals.
How you can implement it
Tools and review intervals are examples; choose them according to risk, obligations and your operating environment.
Map each business use case to a masking method based on sensitivity, downstream need, and reversibility. For software testing, prefer tokenisation or deterministic masking for stable joins; for analytics, strip direct identifiers and generalise quasi-identifiers; for published data, confirm true anonymisation only after re-identification checks and approval by the data owner.
Document field-level rules in the Data Masking Procedure, including format-preserving substitutions, substitution dictionaries, date shifting, nulling of rare values, and exception handling for support cases. Assign accountability to the application owner for implementation, the data steward for review, and the privacy lead for personal-data decisions.
Control extracts, refreshes, and ad hoc copies through approved pipelines rather than manual spreadsheets. Require request tickets to state purpose, dataset, masking rule set, expiry date, and recipient. If a masked file must be shared externally, record the disclosure basis and confirm that the recipient cannot reverse the transformation without keys or reference tables.
Validate masking with test cases that attempt joins, pattern searches, and small-sample deanonymisation. Keep evidence of review results, failed re-identification attempts, and issues corrected in the next release. Use these results to adjust rules when new fields, new integrations, or merged datasets change the exposure profile.
What an auditor will ask to see
Masking matrix showing field, method, rationale, exceptions, and approval owner.
Before-and-after sample extract or test data screenshots with direct identifiers removed or transformed.
Validation notes from re-identification testing, including failed join attempts and approved remediation items.
Baseline documents in the kit
These free kit documents help you document or evidence A.8.11. Each one is tailored to your organization when you build your kit.
Use the SIG and CAIQ questionnaire accelerator to find supporting documents. A mapping is a starting point: verify your actual operating evidence before answering a buyer.
For help adapting this to your environment, explore ISO/IEC 27001 consulting, starting with a free one-hour consultation.
Get these documents tailored to your organization
109 ready-to-edit documents, including all 19 mandatory ones, tailored from a short scoping questionnaire and linked to your risk register and SoA.