Of two pieces of evidence prepared for the same control, one convinces the auditor at a glance, the other leads to more questions and a request for a live demonstration. The difference is usually not in the content but in the type of evidence and how well it explains itself. A screenshot with no visible date, server name or filter is a claim in the auditor’s eyes. Below we compare evidence types for reliability and reproducibility, and explain the elements of a valid screenshot, the integrity of exported reports and the masking of evidence containing personal data.
Evidence types and reliability ranking
ISO 19011:2018 lists the evidence-based approach among its audit principles: audit conclusions must be based on verifiable information. ISA 500, used in financial audit, gives similar general principles: evidence from an independent source is more reliable than internal evidence, evidence obtained directly by the auditor more reliable than indirect evidence, and documentary evidence more reliable than oral statements. Applying these principles to the typical evidence types in ISMS audits gives this table:
| Evidence type | Reliability | Reproducibility | Typical use |
|---|---|---|---|
| System report / export | High | High (with the same query) | User list, backup job history |
| Configuration output | High | High | Password policy, SSH setting, firewall rule |
| System-generated record | High | Intermediate | Log, workflow approval, ticket history |
| Manually kept record | Intermediate | Low | Visitor book, Excel tracking list |
| Screenshot | Medium–low | Low | Settings that can’t be reported |
| Observation | High (for that moment) | None | Clean desk, door control |
| Interview | Supporting | None | Awareness, process knowledge |
An interview alone doesn’t evidence a control; it confirms or calls into question other evidence. Decide in advance in the evidence matrix which type will be the primary evidence for which control. There’s no debating types on audit day.
How are observations and interviews recorded?
Observations and interviews are evidence the auditor obtains themselves; they’re not something the organization provides as a file. Still, the organization can make it easier for this evidence to form. Keep ready the records behind controls observed on the site walk (locked cabinet, escorted visitor, screen lock); when the auditor can match an observation to a record, the risk of a finding drops. In an interview, the process the employee describes must be consistent with the written procedure and the system record. A contradiction, such as the procedure requiring two approvals while the employee says they process with one, is where the auditor digs deepest. Do the same three-way comparison yourself in internal audit; you’ll catch the contradiction before the auditor does.
The limits of screenshots
Screenshots are the most used evidence because they’re quick and can be taken on any system. They have four structural weaknesses:
- They’re instantaneous: They show the state of a setting at that second; they don’t show it was like that throughout the period. Not enough on their own for operating effectiveness.
- They lack context: Which server, which tenant, which filter it was taken with is often not visible.
- Their scope is unclear: Only the first page of a list that needs scrolling is visible.
- They’re easy to change: An image can be edited without a trace; the auditor knows this.
For the distinction between design and operating effectiveness, see the article control effectiveness testing. The practical upshot: a screenshot usually shows design. Showing operation needs records repeated within the period.
Elements of a valid screenshot
If a screenshot can’t be avoided, take it so it’s understandable on its own. A screenshot that raises no questions in the audit carries:
- Date and time: Preferably the application’s own time information; if not, the taskbar clock should be visible. The reliability of the clock rests on control A.8.17 (Clock synchronization) in ISO 27001 Annex A.
- Source: The address bar, server name, tenant name or device name should be visible.
- Identity: If the signed-in user is visible, who took it is also clear.
- Filter and parameters: If the list is filtered, the filter field should stay on screen.
- Uncropped window: Crops that cut off context raise suspicion.
- Descriptive file name and note: A name carrying the control, system, date and the person who took it; for example
A8.13_Backup_SRV-FS01_2026-10-14_mkaya.png.
Advantages of system reports and configuration output
Configuration output is both more complete and more reproducible than a screenshot: the auditor verifies the result by asking for the same command to be run in a live session. Adding the system name and date at the top of the output also solves the context problem of screenshots.
# Linux: the SSH server's effective settings, with system name and UTC date
{ hostname; date -u +%Y-%m-%dT%H:%M:%SZ; sudo sshd -T | grep -Ei '^(passwordauthentication|permitrootlogin|pubkeyauthentication) '; } > sshd_evidence.txt
sha256sum sshd_evidence.txt
# Windows: domain password policy and audit policy
Get-ADDefaultDomainPasswordPolicy | Out-File .\password_policy.txt
auditpol /get /category:* > .\audit_policy.txt
Get-FileHash .\password_policy.txt, .\audit_policy.txt -Algorithm SHA256
# Cisco IOS: VTY line configuration
show running-config | section line vty
sshd -T shows not the configuration file but the settings the server actually applies, including includes and defaults. That’s why it’s stronger evidence than the file itself. A situation you may come across: sshd_config says PasswordAuthentication no, but a file under sshd_config.d overrides it; a screenshot of the file gives wrong information, sshd -T doesn’t. The command needs root privileges; if the configuration has Match blocks, run it with the -C user=…,host=…,addr=… parameter to see the effective setting for a specific user or source. By the same logic, the backup software’s job history report proves much more than a single successful job screen; this is covered in detail in the article backup and restore audit evidence.
Evidence integrity: date, source, immutability
Integrity means being able to show that evidence hasn’t changed from the moment it was created until it’s presented to the audit. Measures can be taken at increasing levels depending on the importance of the control:
- A repository that keeps version history and access logs: Enough for most evidence; on platforms such as SharePoint, version history and retention labels do this job.
- Cryptographic hash: The SHA-256 value taken when the file is created is written to a separate record; later the same calculation shows the file hasn’t changed. Get-FileHash is used on Windows and
sha256sumon Linux. - Timestamp: To prove with an independent authority that a file existed at a certain moment, the RFC 3161 timestamp protocol is used. In Türkiye, electronic certificate service providers under Electronic Signature Law No. 5070 offer timestamping services.
Storing the hash value in the same folder and with the same permissions as the evidence is a useless measure: anyone who can change the file can change the hash too. Keep the hash record in a separate location or in a ticket. For the structure and permissions of the evidence repository, our audit evidence management with SharePoint service is relevant, and for central collection of log-based evidence, our logging, monitoring and audit evidence management service.
Masking evidence that contains personal data
User lists, access records, visitor books and incident records contain personal data. KVKK’s (Law No. 6698) principle that data be relevant, limited and proportionate to the purpose of processing requires that the auditor not be shown more data than needed to verify the control. Practical rules:
- Don’t change the original evidence; produce the masked copy separately and record which original it derives from.
- Reduce columns first: if the national ID number or phone number isn’t needed for the control, remove it.
- Partial masking keeps records unique (for example the last four digits of the staff ID).
- Masking mustn’t destroy the evidence’s value; in an access review, the person–role relationship must stay visible.
- Drawing a black box over text in a PDF doesn’t delete the text; the underlying text can be copied. Use a tool with a real redaction feature, or remove the data at its source.
The last point is where people get burned most in the field. A black box isn’t masking.
Decision table for choosing evidence type
| Control question | Primary evidence | Supporting evidence |
|---|---|---|
| Is the password policy configured? | Configuration output | GPO report, screenshot |
| Were backups taken throughout the period? | Job history report | Restore test record |
| Were access rights reviewed? | Workflow record and source report | Removal requests |
| Are logs collected centrally? | SIEM query result | Source list, retention setting |
| Are visitors recorded? | Visitor record | Observation |
| Do staff know how to report incidents? | Interview | Training record, incident report records |
How to build the evidence package for an access review is explained separately in the access review evidence guide. The general rule is simple: primary evidence must be reproducible; screenshots should be used only for information that can’t be reported, and in a form that’s understandable on its own.
Frequently asked questions
Does the auditor have to accept a screenshot?
No. The auditor decides with professional judgment whether evidence is sufficient and reliable. A screenshot without date, source and context may be rejected, or the auditor may ask for the same information to be shown in the live system. Where possible, support the screenshot with a report or configuration output; the chance of acceptance goes up.
Does every evidence file need a hash or timestamp?
No; decide based on the criticality of the control. For most evidence, a document repository that keeps version history and access logs provides enough integrity. A hash value is useful where you need to show a report didn’t change after it was sent for review, such as an access review; a qualified timestamp is useful for records with a risk of legal dispute.
How should I show the auditor a report containing personal data?
Keep the original unchanged; give the auditor a masked copy containing the columns and rows needed to verify the control. Link the masked copy to the original it was produced from by file name and a note. If the auditor wants to verify a single record, the relevant row can be shown from the original in a controlled way.
Are manually kept Excel records considered weak evidence?
They aren’t invalid in themselves, but because they’re easy to change, their reliability is lower than a system record. Keeping the file in a repository with version history, having the date and the person who made the entry on each row and, where possible, comparing against the source system on a sample basis increases its evidential value. To track the source, version and permissions of evidence together, an organized document repository can be set up with SharePoint document and revision management.