Pentest for SOC 2: what the auditor asks for, and what external testing covers
· 8 min read · SOC 2, Pentest reports
If your first SOC 2 audit is on the calendar, there is probably a line on the readiness checklist about penetration testing or vulnerability scanning that nobody on the team owns yet. This article explains what the auditor is actually looking for, where an external test fits, and where it does not.
SOC 2 does not say "you must have a pentest"
SOC 2 is built on the AICPA's Trust Services Criteria. The criteria describe outcomes, not tools. None of them says "commission an annual penetration test".
Penetration testing and vulnerability scanning do appear in the points of focus, the illustrative guidance under each criterion:
- Under CC4.1 (the entity evaluates whether its controls are present and functioning), penetration testing is listed among the kinds of separate evaluation an entity might use.
- Under CC7.1 (the entity detects new vulnerabilities and configuration changes), the guidance mentions vulnerability scans run periodically and after significant changes.
Points of focus are not requirements, but auditors use them. In practice most auditors expect some evidence that you test the systems you expose and fix what you find. A pentest report is the most common way to show it.
What the auditor is looking for
Auditors are not security testers. They are assessing whether a control exists and operates. For external testing, that usually comes down to four questions.
- Did testing happen? A dated report from a test that actually ran.
- Does it cover the system in scope? The assets in the report match the system described in your SOC 2 scope.
- Were findings acted on? Each finding has a status, and the fixed ones have evidence of the fix.
- Is it ongoing? For a Type II audit, which covers a period of time, a single test from before the period is weak evidence. Testing that recurs through the period is stronger.
The fourth point is where many first-time teams are caught out. A one-off PDF from last year shows that testing happened once. It does not show a control operating across an audit window.
What external testing evidence covers
An external test looks at what is reachable from the internet: your domains, APIs, exposed services, TLS and DNS configuration, and the application itself where the test sends real payloads. From a SecQon report, an auditor gets:
- Dated, evidence-backed findings. Every finding carries the request that produced it, and every report is dated. Where a re-test reproduced a finding, the transcript is printed beside it.
- A SOC 2 mapping on every report. Each finding carries an OWASP Top 10 category, and the report maps those categories to trust services criteria. Access-control findings map to CC6.1, for example, and outdated-component findings to CC7.1 and CC6.8. The mapping is at category level and deliberately under-claims.
- Gaps named on the face of the report. Each framework table splits controls into evidenced, evidenced only in part, and not evidenced.
- A record of testing over time. On paid monitoring plans, monitored assets get a daily non-intrusive re-test and a full re-audit every month, so the record shows testing as an ongoing activity.
What it cannot cover
Much of SOC 2 concerns things no one outside your company can see. An external test, from us or anyone else, cannot evidence:
- encryption at rest, backups, or data retention;
- background checks, security training, or access reviews;
- incident response, or internal logging and monitoring;
- your development lifecycle, or your subprocessors;
- your MFA policy, or your remediation policy.
Your auditor will ask you for evidence of each of these directly. Two OWASP Top 10 categories are also outside any external test's reach: Insecure Design (A04) and Security Logging and Monitoring Failures (A09). A SecQon report names both rather than leaving them out.
Automated or human-led?
Whether an auditor accepts an automated external test for a given control is their decision, not ours. Many accept it as evidence of a testing programme. Some, or some of your customers, specify a human-led test.
SecQon is automated, and every report says so. It does not test business logic or race conditions, and it does not chain findings into attack paths. If a contract or your auditor requires a human tester's sign-off, you need one. Many teams run automated testing continuously and still commission a human-led test once a year. Ask your auditor early which they expect. It is a two-line email that can save you buying the wrong thing.
SecQon is not a SOC 2 audit
To be plain about it: SecQon does not audit you, does not issue a SOC 2 report, and does not make you compliant. Only an independent auditor can issue a SOC 2 report. What SecQon gives you is evidence for one part of the picture, external vulnerability testing, with its limits stated.
A practical sequence
- Confirm with your auditor what external testing evidence they expect, and whether automated testing is acceptable.
- Put the internet-facing assets that sit inside your SOC 2 scope under regular testing, ideally before the audit period starts.
- Fix what the first test finds, and keep the re-test that shows each fix.
- Keep the dated reports from across the period. They are the evidence that the control operated.
If you need one report now, the one-off pentest report runs the full Deep and penetration-test scan once. If you need testing that recurs through an audit window, monitoring re-tests daily and re-audits monthly. The SOC 2 readiness page sets out the mapping and its limits in full.