Blog · · 3 min read
Scoping a penetration test an auditor can actually use
A test commissioned for compliance evidence succeeds or fails at scoping. Decide what the evidence has to show before anyone starts testing, and the report will line up with the question being asked.
Red Cell · compliance evidence · scoping · reporting
Many penetration tests are bought because someone outside the company asked for one: an auditor, a customer’s security review, a prospective investor. The test gets done, the report arrives, and then the questions start. Does it cover the systems in scope for the audit? Does it show the findings were fixed? Is it dated inside the period being reviewed? Most of those questions can be answered before testing begins, at scoping.
Start from the question being asked
Before choosing targets or techniques, write down who will read the report and what they need it to show. An auditor reviewing a control, a customer’s security questionnaire and a board asking “could someone get in?” are three different readers with three different questions.
Testing can support compliance evidence and customer assurance requirements. Scope and evidence mapping depend on the applicable requirements. That is why the requirement comes first: it tells you which systems must be in scope, how recent the test must be, and what the report must record for it to be accepted.
Draw the scope around the system that is being assessed
The most common gap between a test and an audit is a mismatch in boundaries. The audit covers the production platform and the services that support it; the test covered the marketing site and one application. Neither side is wrong, but the report cannot be used as evidence for systems it did not touch.
A useful scope statement names:
- The environments in scope — production, staging, or both — and why.
- The asset categories: external perimeter, applications and APIs, cloud accounts, identity, and any AI assistants with access to data.
- What is explicitly excluded, so the exclusion is a decision rather than an omission.
- The testing window, so the report’s dates can be placed against the audit period.
Agree the evidence before the testing
A reviewer reading a report is looking for more than a list of issues. They want to see that the test was authorised, that it was scoped deliberately, that findings were ranked consistently, and that the important ones were addressed. Each of those needs something in the report:
- A record of the authorisation and rules of engagement, with dates.
- The scope statement, matching the one agreed at the start.
- A severity method that is explained once and applied to every finding, rather than ratings that appear from nowhere.
- For each finding, the affected system, the impact, and the recommended fix.
- Retest results for the findings the agreement covers, with the retest date.
If the requirement being evidenced is known, the report can also map findings to it. That mapping is agreed in the statement of work, so both sides know what it will and will not claim.
Plan the retest at the same time
A report full of open findings is evidence that testing happened. A report with retest results is evidence that the programme works. Put the retest into the original agreement: which findings it covers, how long the client has to remediate, and what the retest will confirm. It is far easier to schedule at the start than to negotiate once the findings are in.
In the payments platform engagement, a retest 30 days later confirmed every critical and high finding closed. The client passed its SOC 2 Type II audit, with the engagement report cited as evidence of a functioning security testing programme. The audit and its result belong to the client; the retest is what let the report say more than “these issues existed”.
Keep it a test, not a formality
The risk with evidence-driven testing is that the scope shrinks to whatever is easiest to pass. A test designed only to produce a clean report will produce one, and it will tell you very little. The better approach is to scope for the evidence you need and for the attack paths that matter, and to let the report show both.
If you are preparing for a review and want the scope worked out before testing starts, start with a scoping conversation.