Cookie settings

Necessary, always on: they keep the site working and remember your cookie choice.

Skip to content
Skip to content

Glossary. Audit evidence.

What is audit evidence for data security?

Evidence is a record someone else can re-perform. It needs five fields: who produced it, when, from which system, how it is complete and which control.

By Dipal Panchal, co-founder and CEO · Updated November 8, 2026

A record an auditor can run again

Audit evidence for data security is a record that lets an auditor re-perform a check on your data estate: who produced it, when, from which system and with what result. Unlike a compliance report, which is the organization’s claim that a control works, evidence is the record the auditor tests that claim against. An item missing its producer, date, system, completeness check or control is a claim, however neat it looks.

NIST defines evidence as “grounds for belief or disbelief”. For data security, the grounds are records the systems produced, with dates.

An evidence register, two items rejected

A team’s register for one audit period on one estate, not IRIS output. The auditor decides acceptance; Status says whether each item passes the five-field test and what it supports.

Records
SourceStatusReasonOwner
Access list export, billing databasePasses; supports PCI DSS v4.0.1 7.2.4Run November 3, 2026 by an account that sees every grant; query attached; 214 effective grants traced to rolesPayments engineering
Coverage records, four monthly scans to October 28, 2026Passes; supports ISO/IEC 27001:2022 A.5.9Every source in the asset inventory has a status or a recorded exclusion; the failed one has an owner and due dateData platform lead
Access review decisions, third-quarter cyclePasses; supports SOC 2 (2017 criteria) CC6.3List exported at cycle start, reconciled to accounts reviewed; reviewer, reason and date on every decisionSecurity operations
Scans before and after the crm.contact fixPasses; supports ISO/IEC 27001:2022 A.8.11Both dated, same scope, linked to change record CHG-0412; the after scan flags sampled values as maskedCRM owner
Screenshot of the admin console’s user listRejected: fails the five-field testNo date, no system name and no way to tell what the screen filtered outIT operations
Closed ticket: “contractor access removed”Rejected as proof of removalThe ticket says the work was done; the proof is the scan that found the grant and the November 4 scan that doesn’tService desk

The five fields every item needs

Sylox’s test: an item passes when it carries five things: who produced it, the date, the system it came from, how completeness is shown and the control it supports. In financial-statement audits, PCAOB AS 1105, paragraph .10, and ISA 500, paragraph 9, direct auditors who use information the entity produced to evaluate whether it is reliable, including whether it is complete and correct.

Row 1 carries the query, the account that ran it and a count traced to role memberships. An export with no query attached can’t be shown to be complete, whatever it contains.

Why the screenshot and the ticket fail

Row 5 has nothing an auditor can re-perform: no date, no system and no record of what the screen hid. Row 6 is a statement that work happened. The scan of November 4, which no longer finds the contractor’s grant, is the record that shows it.

The same test rules out a spreadsheet with no provenance and a tool score with no findings behind it. A single point-in-time scan offered for a whole period fails too, unless scans repeat across the period.

Where the evidence comes from today

Dipal Panchal, Sylox’s co-founder and CEO, describes one audit flow he has seen in the security teams he has worked with. The CISO forwards each audit question to every application owner, who answers by hand. The answers come back by email, and the auditor files and labels them in a separate folder.

Each manual step can drop a field: an answer typed from an owner’s knowledge may name no system, and an email may carry no date. On where the judgment sits, Panchal is direct: “Are we an auditor? Can we certify whether that evidence is correct? That’s the auditor’s job.”

Sufficiency stays with the auditor

Evidence doesn’t decide whether a control is met; your auditor does, and judges how much evidence is sufficient. Its shape has to match the report. A point-in-time report needs evidence as of a date, and a period report, such as a SOC 2 Type 2, needs evidence from across the period.

A configured rule is evidence of design: it shows the control exists. Evidence that it operated is dated records from across the period.

With IRIS

IRIS generates evidence items from scan data and collects them in a register you can export as CSV or XLSX. Capabilities vary by edition.

The findings behind each item, with their source and scan date: IRIS produces part of this, for items generated from scan data.

What each scan read and couldn’t read: IRIS produces part of this, for the tables and files the scan account can list.

The control each item supports: IRIS produces part of this, for the frameworks IRIS maps; for ISO/IEC 27001, SOC 2 and PCI DSS, evidence that supports your assessment.

Whether the evidence is sufficient, or a control is met: Not produced by IRIS.

Explore audit evidence

Re-perform one item before the auditor does

Pick one item from your last audit and check its five fields. Run its query again with the same account and confirm the count, then compare it with a run that doesn’t share that account’s blind spot, such as an administrator’s.

For scan-based items, the record behind them is scan coverage. To prepare a full set, work through the evidence readiness checklist.

FAQ

What compliance teams ask

Protect your brand. Start with your data.

See a sample report, then walk through the evidence behind it with us.