The short answer
ISO 27001 is a management system standard. It tells you to identify your information security risks and choose controls to treat them, then check that those controls work. It doesn’t prescribe specific technical activities, and penetration testing isn’t in the clauses or in Annex A as a named requirement.
What the standard does require is that you can find and deal with technical weaknesses in the systems inside your scope, and that security is tested when you build or accept systems. For a business running a web application, an API or anything else reachable from the internet, a penetration test is the clearest and most widely accepted way to show that. So the honest answer is: not mandatory, but expected for most technology businesses, and you need a good reason if you choose not to.
The controls a pen test supports
Several Annex A controls are relevant. Titles below are from ISO/IEC 27002:2022, which gives the implementation guidance for each one.
| Control | What it asks for | Where a pen test fits |
|---|---|---|
| 8.8 Management of technical vulnerabilities | Get information about technical vulnerabilities in your systems, evaluate your exposure and take appropriate action | Finds exploitable weaknesses a scanner misses; findings feed your vulnerability and patching process |
| 8.29 Security testing in development and acceptance | Define and run security testing as part of the development life cycle and before accepting new or changed systems | A test before go-live or after a major release is acceptance testing with an attacker’s view |
| 8.25 to 8.28 Secure development life cycle, application security requirements, secure system architecture and engineering principles, secure coding | Build security into how you design, specify and write software | Tests whether those practices actually produced a secure application |
| 5.35 Independent review of information security | Review your approach to information security independently, at planned intervals and after significant change | An independent tester’s results are useful input, though they don’t replace a review of the ISMS itself |
| 5.36 Compliance with policies, rules and standards for information security | Regularly check that your own policies and standards are followed, technical ones included | Shows whether hardening, access control and configuration standards hold up in practice |
| 8.34 Protection of information systems during audit testing | Plan and agree any testing of operational systems so it doesn’t disrupt them | Rules of engagement, test windows and approvals for the pen test itself |
The ISO/IEC 27002 guidance for vulnerability management and security testing mentions penetration testing as one of the techniques you can use. That guidance isn’t a set of requirements, but auditors know it well and it shapes what they expect to see.
Your risk assessment and SoA make the decision
Whether a pen test is the right way to meet these controls is decided by your own ISMS, not by a checklist. Clause 6.1.3 requires you to choose the controls needed to treat your risks, compare them against Annex A and record the result in your Statement of Applicability, with the reason each control is included or excluded.
Work it through from the risks. If your risk assessment identifies unauthorised access to customer data through your web application, your treatment needs to include a way of finding the weaknesses an attacker would use. Automated scanning covers known vulnerabilities in software and infrastructure. It doesn’t show whether one customer can read another’s records, or whether your authentication can be bypassed. A pen test does.
Two consequences follow. First, once your SoA or your vulnerability management procedure says you pen test annually, that becomes an audit criterion: the auditor will check you did it, and a missed test is a finding. Second, if you decide not to pen test, the SoA and risk treatment plan should show how else you’re detecting exploitable weaknesses. For an organisation with no internet-facing applications and a small, managed SaaS footprint, regular scanning and configuration review can be a defensible answer. For a SaaS company, it rarely is.
Why auditors and customers expect one anyway
Certification auditors can’t demand a pen test the standard doesn’t require. What they can do is ask how you know your public-facing systems are secure, and test whether your answer holds up. For a business whose product is a web application, “we run a scanner” leads to follow-up questions about access control, authentication and the custom code no scanner understands. An independent pen test answers those questions directly.
Customers are more direct. Enterprise security questionnaires routinely ask when your last independent penetration test was, who performed it, and whether high-risk findings were fixed. Many ask for an executive summary or attestation letter. For SaaS providers in particular, the pen test sits alongside the ISO 27001 certificate as standard evidence in procurement, and its absence slows deals down. Our guide to ISO 27001 for SaaS companies covers the wider picture.
Scanning is not a pen test
The two are often confused. A vulnerability scan is automated and frequent, and finds known issues such as missing patches and misconfigurations. A penetration test is manual, goal-driven and periodic: a skilled tester tries to exploit weaknesses, chain them together and show the real impact. Control 8.8 needs both kinds of activity in most environments: scanning for breadth and frequency, testing for depth. The full comparison is in penetration test vs vulnerability scan.
When to test in the certification cycle
Timing matters more than most teams expect. The best point for the first test is after your main controls are in place but early enough before the Stage 2 audit to fix the serious findings and retest. That lets you show the auditor the full cycle: test, assess, treat, verify. A report that lands days before Stage 2 with open critical findings proves you have problems, not that you manage them.
After certification, set the frequency in your own procedure. Common practice, and what most customers ask for, is at least once a year, plus a test after significant change: a major release, new infrastructure, a cloud migration or an acquisition. That lines up with Clause 8.2, which requires you to reassess risk when significant changes happen, and with 8.29 for acceptance of changed systems. Plan annual tests so a recent report is available for each surveillance audit in the three-year audit cycle.
Choosing the scope
The pen test scope should follow your ISMS scope and your risks. Test the systems that hold the information you’re certifying, and start with what an attacker can reach.
| Test type | What it covers | When it matters most |
|---|---|---|
| External infrastructure | Internet-facing hosts, services, DNS, email and exposed admin interfaces | Everyone with a public footprint |
| Web application | Authentication, sessions, access control, injection and business logic, ideally as a logged-in user (grey-box) | SaaS products and customer portals |
| API | Authorisation on every endpoint, input handling, rate limiting and data exposure | Mobile back ends, integrations and API-first products |
| Cloud configuration review | Identity and access, storage exposure, network rules and logging in AWS, Azure or Google Cloud | Cloud-hosted platforms, especially after migration |
| Internal network | What an attacker or compromised device inside the network could reach | Offices with on-premises servers or flat networks |
Use a recognised methodology. NIST SP 800-115 is a widely used guide to planning and running security testing, and the OWASP Web Security Testing Guide is the reference for web applications. Agree rules of engagement, test windows and who to call if something goes wrong before testing starts; that is control 8.34 in practice.
What evidence auditors want
The report alone is half the story. The auditor is checking that the test fed into your management system:
- The report, or at least its executive summary and findings list, showing scope, dates, methodology and who performed it. Some auditors will view a full report on screen rather than take a copy.
- Triage of each finding, rated against your own risk criteria and recorded in your risk register or vulnerability tracker with an owner and a due date.
- Treatment: tickets, changes or configuration fixes showing what was done, plus documented risk acceptance by the right owner for anything you chose not to fix.
- A retest confirming that critical and high findings were actually resolved.
- Lessons applied: where a finding revealed a weakness in a process, such as code review or change management, a corrective action under Clause 10.2.
Common mistakes are a scope that leaves out the main product, a test from two years ago, findings with no owners, and remediation dates that quietly passed. Each of those turns good evidence into an audit finding.
What drives pen-test effort
Pen tests are scoped by effort, and the main drivers are predictable: the number of applications, APIs and IP addresses in scope; the number of user roles to test; how complex authentication and business logic are; black-box versus grey-box; whether cloud configuration or internal networks are included; and whether a retest is part of the engagement. A tightly defined scope based on your risks gets you better coverage of what matters than a broad, shallow one. For the full picture of what ISO 27001 involves, see what drives the cost of ISO 27001.
Our penetration testing covers web applications, APIs and external infrastructure, black-box or grey-box, with reports written for both your engineers and your auditor. Linking findings to risk treatment is the part we take on for clients, so the test becomes audit evidence rather than a document in a drawer.
SOC 2, for contrast
SOC 2 works the same way in principle. The AICPA’s Trust Services Criteria don’t require a penetration test, but penetration testing is listed among the points of focus for monitoring controls, and CPA firms commonly ask for one as evidence that your controls are evaluated. The difference is in how it’s reported: in a SOC 2 report, the auditor describes the controls you have and how they were tested, so readers can see whether pen testing is one of them. Customers read that closely. For how the two frameworks compare overall, see ISO 27001 vs SOC 2.
Next step
If you’re building your ISMS yourselves, check that your SoA, vulnerability management procedure and pen-test plan say the same thing, then test the evidence before the certification body does. ISO 27001 requires internal audits to be objective and impartial: staff can do it if they aren’t auditing their own work, which is why many small teams bring in an outside auditor. Our independent internal audit is run by a Lead Auditor the way the certification body will audit you. To have the ISMS built with you, see our ISO 27001 service, or book a free 30-minute consultation to talk through your testing scope.
