What a pentest report needs to get through a customer's security questionnaire
· 7 min read · Pentest reports, Security questionnaires
A prospective customer has sent a security questionnaire, and somewhere in it is a line like "Provide your most recent penetration test report" or "When was your last external security test, and what were the results?" You attach a PDF and move on.
The person on the other side is usually a security analyst working through a pile of vendor reviews. They are not reading your report for interest. They are checking whether it answers their questions, and whether it raises new ones. A report that looks impressive but leaves them guessing about scope or dates comes back as a follow-up email, and the deal waits.
This is the checklist we would run against any report before attaching it, including one of ours.
1. It is dated, and the date is recent
The first thing a reviewer checks is when the test ran. Many questionnaires ask for a test within the last twelve months; some ask for the last quarter.
- The report states the date the test ran, not only the date the PDF was generated.
- If findings were fixed afterwards, the report or a follow-up record says when.
- A report shared through a link should be a dated document, not a live page that changes after you send it.
2. It says exactly what was tested
"We tested our application" is not a scope. The reviewer wants to know which hosts, which surfaces, and from where.
- Named assets: the domains, subdomains, IP addresses or API base URLs that were in scope.
- Whether the test ran from the internet (external) or from inside your network.
- Whether pages behind your login were tested, or only what an anonymous visitor can reach.
- Whether subdomains were tested, or only the apex domain.
If the questionnaire asks about your production application and the report covers only your marketing site, the reviewer will notice.
3. Each finding carries its evidence
This is the line between a pentest report and a scanner export. A finding without evidence is a claim the reviewer has to take on trust, and they will not.
For every finding, look for:
- the request that produced it, and the part of the response that showed the problem;
- a severity, and a CVSS score where the source publishes one;
- how far it was proven, not only what a tool reported (see the next point);
- a plain-language explanation of the impact;
- specific steps to fix it.
4. It separates what was proven from what was suspected
A report with 300 findings and no distinction between them tells the reviewer nothing about your real exposure. A report that labels each finding by the strength of its proof tells them a great deal.
In a SecQon report every finding carries one of four labels: demonstrated when a bounded proof-of-exploit changed how the application behaved, confirmed when a re-test observed the same thing again, likely when it was seen once and could not be reproduced, and unconfirmed when only a tool reported it. Whatever vocabulary a report uses, the reviewer should be able to tell a reproduced issue from a tool's guess. We wrote a longer guide on how to read those labels.
5. It maps findings to something the reviewer recognises
Security teams think in frameworks. A finding labelled "Missing Content-Security-Policy header" means more to them when it also says which OWASP Top 10 category it falls under, and which control in the framework they care about it bears on.
- OWASP Top 10 categories per finding.
- A mapping to the framework the buyer named, often SOC 2 trust services criteria or ISO 27001 controls.
- A mapping that under-claims: a finding should map only to controls its category actually supports.
6. It states its own limits
This is the item most reports skip, and the one experienced reviewers trust most. Every test has limits. A report that names them reads as honest. One that implies it covered everything reads as either naive or misleading.
- Automated or human-led, stated plainly.
- What was not tested: for example business logic, race conditions, or chained attacks.
- The OWASP Top 10 categories that cannot be assessed from outside. Insecure Design (A04) and Security Logging and Monitoring Failures (A09) are not observable from the internet by anyone, and a good external report says so instead of leaving a silence.
7. Open findings have a status
If the report lists a high-severity finding, the reviewer's next question is "is it fixed?" Answer it before they ask.
- Each finding has a status: open, in progress, fixed, or accepted as a risk with a reason.
- Fixed findings have proof of the fix, ideally a re-test that no longer reproduces them.
The questions a report cannot answer
A pentest report answers the external-testing questions in a questionnaire, and only those. A typical questionnaire also asks about encryption at rest, background checks, security training, access reviews, backups, incident response, internal logging, secure development, subprocessors, data retention and MFA policy. No external test can see any of these. You answer them from your own policies and records.
SecQon's Security Questionnaire Pack is built around this split. It re-presents one scan as 42 answers in the style of the SIG Lite and CAIQ questionnaires. Each question is answered from scan evidence, marked as not externally assessable, or marked as not assessed with what would assess it. Fourteen common questions are marked not externally assessable, with a note on who in your company can answer them. There is more on how the pack works on the security questionnaire page.
Before you attach it
Run the seven points above against the report you are about to send. If any of them fails, fix the gap before you attach it. That might mean a covering note on scope, an updated status column, or a fresh test. If you do not have a report yet, the sample SecQon report shows each of these sections on a real scan.