For teams scanning mailboxes and shared folders in audit week, the problem usually isn’t missing evidence. The real problem is that it isn’t written down anywhere which control is met by which record. An evidence matrix closes this gap: for each control, it keeps in a single table what the evidence is, who produces it, how often it’s renewed and where it’s kept. The model below is built tool-independently, column by column, for ISMS managers and teams preparing for TISAX.
What does an evidence matrix solve?
The Statement of Applicability (SoA) says which controls are implemented and why; the risk treatment plan explains the reason for implementing them. Neither answers the question “which record do we use to show this control works?”. The matrix is this third layer and solves three problems you may come across:
- Evidence tied to a person: If only the system administrator knows the backup evidence, the audit session stalls while that person is on leave. The matrix ties evidence to a role.
- The same work done twice: In an organization running ISO 27001 and TISAX together, the same access review report sits in two folders, in two different versions. The matrix makes a single source and multiple mappings visible.
- Periodic evidence slipping: If a review that should be done quarterly is noticed only before the audit, people try to produce it retroactively. That’s the first thing the auditor catches. The frequency and last-produced-date columns show this risk months in advance.
The matrix isn’t a copy of the SoA. A control that is one row in the SoA can split into several evidence rows in the matrix, and a single evidence row can be linked to several controls. That’s why the matrix’s primary key is not the control but the evidence item. This decision shapes everything.
Mandatory columns and their definitions
The columns below are set up the same way in a spreadsheet, a SharePoint list, a GRC tool or a ticketing system. Fix the column names within the organization. Columns renamed every audit period make comparison with past periods practically impossible.
| Column | Content | Example value |
|---|---|---|
| Evidence ID | A unique code that never changes | EVD-047 |
| Evidence name | What is shown | Quarterly AD permissions review |
| Mapped controls | Standard and control numbers | ISO 5.18, 8.2; VDA ISA 4.2.1 |
| Evidence type | Record, report, configuration, screenshot, observation | System report + approval record |
| Evidence owner | Role responsible for producing it | System administration team lead |
| Frequency | Event-based, monthly, quarterly, yearly | Quarterly |
| Last produced date | Date of the latest evidence | Last working day of the previous quarter |
| Storage location | Permanent, versioned location | Evidence library / 05-Access |
| Retention period | Period consistent with the records list | Certification cycle + 1 year |
| Status | Current, overdue, missing | Current |
Two more columns are very useful in practice: source system (which system the report was pulled from, with which query or menu) and verifier (a second person who checks the evidence is adequate). The source system column lets the same evidence be reproduced the same way in the next period. When the auditor asks “can you pull this report again now?”, the answer is ready. You open the console and run the query.
How to choose evidence for a control
Reading the control text and asking “is there a document that shows this?” fills the matrix with policy files. A policy only shows the control’s design; the auditor wants to see that the control is operated. Ask these three questions in order for every control:
- Design evidence: Which approved document defines how the control should work? (Policy, procedure, standard configuration.)
- Operating evidence: Which record shows the control actually worked throughout the period? (Review report, ticket record, log, test result.)
- Outcome evidence: Which record shows what was done about the deviations the control found? (Removed permission, closed vulnerability, corrective action record.)
The third question is the one most often skipped. A vulnerability scan report shows the control is operated; but if there’s no record of how quickly critical findings were closed, the auditor questions the control’s effectiveness. Scanning isn’t enough. You need a closure trail. The reliability of the evidence type also drives the choice: a dated report pulled directly from the system with a known source is stronger than a screenshot showing the same information. The details of this difference are in the guide on the reliability difference between screenshots and system reports.
One piece of evidence for several controls: multi-standard mapping
ISO/IEC 27001:2022 Annex A and VDA ISA overlap to a large extent; technical and organizational measures under KVKK and security clauses in customer contracts also often rely on the same records. The logic of multi-mapping is simple: evidence is produced once, kept as a single row in the matrix, and all related requirements are listed in the “mapped controls” column.
Three rules for running this soundly:
- The strictest requirement decides. While ISO 27001 leaves review frequency to the organization, a customer contract or the TISAX target maturity level may require a more frequent or better documented practice. Write the most frequent one in the frequency column.
- Note scope differences. The scope of a TISAX assessment may include locations different from the scope of the ISO 27001 certificate. If a single piece of evidence doesn’t cover both scopes, split the row in two.
- Fix the numbering version. VDA ISA control numbers can change between catalog versions. Write the catalog version you use in the matrix header; recheck the mapping during transitions. What to look at during a version transition is covered in the article comparing TISAX and ISO 27001 shared controls.
Another benefit of multi-mapping is impact analysis. When production of a piece of evidence falters, you can see which standards are affected within seconds by filtering the matrix.
Sample matrix rows: organizational, people and technological controls
The rows below are examples from the ISO/IEC 27001:2022 Annex A themes. VDA ISA numbers follow the ISA 6 catalog; verify them in the catalog version you use.
| Evidence | ISO 27001 Annex A | VDA ISA | Owner / frequency |
|---|---|---|---|
| Approved asset inventory export and ownership list | 5.9 | 1.3.1 | IT asset manager / quarterly |
| Supplier security assessment records | 5.19, 5.21 | 6.1.1 | Purchasing / at contract start and yearly |
| Awareness training completion report and assessment results | 6.3 | 2.1.3 | HR / yearly and on joining |
| Access review report, decisions and removal tickets | 5.18, 8.2 | 4.2.1 | System administration / quarterly |
| Backup job reports and restore test record | 8.13 | 5.2.9 | Infrastructure team / monthly test |
| Change record list and sample approvals | 8.32 | 5.2.1 | Change manager / continuous |
In the backup row, the monthly job report alone isn’t enough. Green “Success” rows show the backup was taken, not that it restores. The restore test’s target, result and duration must also be part of the evidence definition; the content of this set is explained in detail in the backup and restore audit evidence guide.
For people controls, evidence contains personal data. When a training report or pre-employment screening record is linked to the matrix, access permissions on its storage location should be kept narrower than for the rest of the evidence library.
Keeping the matrix current, and ownership
The matrix is at its most accurate on the day it’s built. Without maintenance it goes stale in a few months. A sustainable setup looks like this:
- The matrix has a single owner: Usually the ISMS manager. They don’t produce the evidence; they manage the completeness of rows and the status column.
- Evidence owners are tied to roles: A role is written instead of a person’s name, and the role–person mapping is kept in a separate RACI table. When duties change, the RACI is updated, not the matrix.
- Monthly status scan: If “last produced date + frequency” is earlier than today, the row is automatically marked “overdue”. In a spreadsheet this is simple conditional formatting; in list-based tools it’s a calculated column.
- Trigger events: The matrix is reviewed when a new system goes live, the SoA is revised, the scope changes or the standard version is updated.
- Management review input: The number of overdue and missing evidence items can be reported in the management review as an information security performance indicator.
The platform where evidence lives is independent of the matrix. Still, for the storage location column to work with permanent links, building the document library’s folder structure to match the matrix’s theme classification makes the job much easier. For SharePoint users, the library structure, versioning and permission model are explained in the article evidence management with SharePoint. Teams who want to set up the evidence library and the matrix together can look at our audit evidence collection and SharePoint service. To track the source, version and permissions of evidence together, an organized document repository can be set up with SharePoint document and revision management.
Mistakes you may come across
- Building the matrix one-to-one with the SoA: One row per control, with a policy name in the “evidence” column. This structure never shows operating evidence.
- Leaving the evidence definition vague: Instead of writing “logs”, write a reproducible definition such as “export of the alert rule list from the SIEM and the monthly alert review record”.
- Using the file name instead of an evidence ID: When the file is renamed, the link breaks. The evidence ID doesn’t change; file names carry the ID as a prefix.
- Undated evidence: A report without a production date can’t prove it belongs to the audit period. The date must be inside the report or in the file metadata.
- Bulk production before the audit: Four quarterly reviews for a year all dated the same week is spotted at first glance in sample testing. It’s a sign the control wasn’t operated.
- Duplicating TISAX and ISO evidence in separate folders: The two copies drift apart over time; the auditor meets two different versions. The single-source, multi-mapping rule must be kept. In preparing the evidence file specific to a TISAX assessment, our TISAX pre-assessment preparation work sets up this distinction from the start.
When starting, don’t pour the whole of Annex A into the matrix at once. Start with the five areas that produce the most findings: access management, change management, backup, suppliers and training. Once these rows settle, the remaining controls are added quickly with the same template. For the official text of ISO/IEC 27001:2022, the reference point is ISO’s standard page.
Frequently asked questions
Can the evidence matrix and the SoA be the same document?
They can, but we don’t recommend it. The SoA shows whether a control is implemented and why; the evidence matrix tracks evidence items, their owners, frequencies and storage locations. If you merge them, the SoA has to be revised with every evidence update and document control gets harder.
How much evidence is enough for one control?
Coverage, not count, decides. For every control there should be at least one design evidence item (an approved document) and one operating evidence item showing it was operated throughout the period. For controls that find deviations, outcome evidence showing how the deviations were closed should also be added.
Do we need special GRC software for the evidence matrix?
No. These columns can be set up in a spreadsheet or SharePoint list. What you really need is fixed evidence IDs, a versioned storage location, role-based ownership and status tracking that shows delays.
What happens to the matrix when the VDA ISA catalog version changes?
Evidence rows usually stay the same; what changes are the mapped control numbers and the scope of some requirements. During the transition, the mapping column should be rechecked against the new catalog and the catalog version in the matrix header updated.