Audit strategy

What auditors really look for in evidence collection

Most teams collect documentation, not evidence. Here is what auditors actually grade on and how to keep your program audit-ready all year.

Sarah Jenkins

By Sarah Jenkins

Regulatory & Compliance Analyst · 5 min read · August 5, 2026

When you prepare for a SOC 2 Type II or ISO 27001 audit, the impulse is to collect everything. Screenshots, logs, spreadsheets, policy documents — all of it dumped into a shared folder two weeks before fieldwork begins. The auditor arrives, opens the folder, and the first thing they ask is: "Can you map each piece to a specific control?"

That mapping step is where most evidence programs fall apart — and it is entirely preventable.

The difference between documentation and evidence

Evidence is not documentation. Documentation proves something exists. Evidence proves a control *operated effectively* over a defined period. The auditor is not asking whether you have a change management policy. They are asking whether change management ran correctly for the past twelve months, and whether you can produce a representative sample of changes that followed it.

The difference is operational. A policy document lives in a wiki. Evidence lives in the systems where work actually happened — your ticketing tool, your IdP, your SIEM, your MDM. If your evidence program consists of exporting those systems the week before fieldwork, you are building a case after the fact, and auditors can usually tell.

Auditors distinguish between evidence that was captured at time of operation and evidence that was assembled in anticipation of inspection. Both may document the same event. Only the first carries full weight.

What the auditor is actually testing

Auditors evaluate two properties for every control.

Design effectiveness — does the control, as designed and implemented, address the stated risk? Your policy exists. Your tool is configured. The logical chain from threat to mitigating control is coherent. Operating effectiveness — did the control run the way it was designed to, consistently, over the period under review? This is where evidence lives.

For operating effectiveness, auditors sample. They draw from the population you present — not from examples you select. Your job is to present the complete population. Their job is to pull from it.

This reframes the problem. You are not collecting evidence for an audit. You are maintaining a population of records that proves a control ran. The audit is a sampling exercise on top of that population.

Five attributes auditors grade evidence on

When an auditor reviews a submission, they grade it silently on five dimensions.

Completeness. Does the record capture the full activity? A ticket marked "closed" says nothing about who approved the change, what was reviewed, or whether an exception was logged. A ticket with an approval timestamp, approver name, and attached change form does. Timeliness. Was the evidence created at the time of the activity, or assembled later? Retroactive screenshots and undated approval emails are red flags. Auditors check whether metadata timestamps match document dates. Precision. Does the evidence address the specific control objective? A screenshot of your MDM enrollment page addresses device inventory. A screenshot of your MDM dashboard summary does not — it shows the tool exists, not that the control is running. Integrity. Can the auditor trace the evidence to an authoritative system? Read-only links to your SIEM, ticketing system, or IdP carry more weight than exports. An export asks the auditor to trust you did not filter rows. A live source does not. Attribution. Is it clear who performed the activity, and when? Anonymous actions and shared credentials fail this check. Individual user actions logged in the original system pass it.

Three requests that reliably derail fieldwork

Three evidence requests slow audits down more than any others. All three are predictable and avoidable.

"We need the access review for Q2." The team searches Confluence, finds a spreadsheet, cannot locate the manager approval column, finds an email thread, reconstructs the approval chain, and delivers something four days later. The fix: treat access reviews as a structured workflow with a locked output — one document per quarter, approver records attached, stored in one location. Run a self-test before fieldwork: can you retrieve the Q2 access review in under ten minutes? "Can you show us evidence that vulnerability management ran last month?" Teams point to their scanner. The scanner shows findings. The auditor then asks: "What is your SLA for critical findings, and can you show those findings were closed within that SLA?" Most teams track finding evidence but not SLA-compliance evidence. "We need evidence that your backup restoration test worked." Not that the backup ran — that you *tested restoration*. Backup run logs are not restoration evidence. A dated runbook completion record with a named operator and a recorded result is.

How to run an evidence-ready program year-round

The most effective audit preparation is a quarterly self-assessment using the same sample pull the auditor would use.

Pick five controls. Pull the auditor-equivalent sample yourself. Grade each piece against the five attributes above. If a piece does not pass all five, close the gap before fieldwork begins.

For high-volume controls — access logs, change records, vulnerability scans — the failure mode is usually incomplete populations and missing SLA documentation. For low-frequency controls — risk reviews, vendor assessments, policy attestations — it is usually timeliness: the activity happened, but the evidence was created days or weeks later.

Running this cycle quarterly means your evidence file is production-ready year-round. You know what you have, where it lives, and whether it holds up.

One practical discipline: assign an evidence owner for each control family. That person is accountable for the population, not just the output. When the auditor asks for it, one person raises their hand without hesitation.

Audit prep should not be a six-week scramble. CloudAnzen continuously maps your stack to SOC 2 and ISO 27001 controls, flags evidence gaps in real time, and keeps your program audit-ready every day of the year — not just the weeks before fieldwork begins. Talk to us.

Keep the momentum

Turn this guidance into a working program

CloudAnzen helps teams connect evidence, review failing controls, manage risk, and stay audit-ready across frameworks from one place.