Skip to main content
Security guideReviewed

How to answer a customer security questionnaire with evidence

A larger customer has sent a security questionnaire with a due date.

This guide shows what the common questions are really asking, which of them a penetration test or security review can evidence and which it cannot, and how to word an honest answer when the true answer is "not yet".

Security questionnaire evidence crosswalk

Find the row closest to the question in front of you, check the evidence column against what you actually hold, then adapt the example answer with your own systems, dates and providers. The example answers are written for an illustrative business and qualify the claim wherever a plain yes would say more than the evidence shows.

How to use it

Only use a sentence that is true for you.

17 rows

Security questionnaire evidence crosswalk

Typical questionWhat the reviewer is really askingEvidence that answers itWhere that evidence stopsExample answer that does not overstate
Do you conduct annual penetration testing of your application?Is an outside party testing the system we will use, and is that a repeating programme or a one-off?The report's scope as tested: targets, roles, build and dates. A booking or statement of work for the next test, if one exists.One report shows one test. It does not show a programme, or anything outside its scope."Our customer portal and its API were penetration tested in March 2026 by an external provider. This was our first test. The next is planned for March 2027."
When was your last penetration test, and can you share the results?Is the evidence recent enough to describe the system as it runs today?The report date and version, the executive summary and the findings table.Releases since the test are not covered. A large change since then, such as a new module, sits outside the evidence."Last tested in March 2026, before our billing module launched in June. We can share the executive summary and findings table under a confidentiality agreement. Billing is in scope for the next test."
Were any critical or high-risk findings identified, and have they been remediated?Is a known serious weakness still open in the system we will rely on?The findings table with business-risk ratings and CVSS v3.1 scores, the remediation plan, and the retest section marking each finding fixed, partially fixed, still reproducible or not retested.Fixed means fixed in the retested build. A finding marked not retested is unverified, whatever the developer reports."The test reported one critical and two high findings. The May retest recorded the critical finding and one high finding as fixed. The other is partially fixed; the remaining change ships in July and will be retested then."
Was the test performed by an independent third party?Could the tester have been marking their own work?The provider's written statement of whether it built, hosts or maintains the system. A Clever Ops report records who built and who hosts a system when that is Clever Ops.A review by the builder or host is useful for fixing problems but is not independent evidence."The March review was carried out by the agency that built and hosts the application, and its report says so. It was not an independent test. An independent test by a separate provider is booked for August."
What methodology or standard was the test performed against?Was the testing structured, or was someone clicking around the site?The report's methodology section naming its references, such as the OWASP Web Security Testing Guide and ASVS 5.0, and the OWASP Top 10:2025 category on each finding.Referencing a framework describes how the work was organised. It is not a certification against that framework."Testing referenced the OWASP Web Security Testing Guide and ASVS 5.0, and each finding is mapped to an OWASP Top 10:2025 category. The test was not a certification against those documents."
Is your application protected against the OWASP Top 10?Has anyone checked for the common categories of web application risk?Findings mapped to OWASP Top 10:2025 categories, plus the negative results for checks that found nothing.The OWASP Top 10:2025 is an awareness document on the most critical security risks to web applications, not a pass mark. A test cannot show that no issue in those categories exists."Our March test of the customer portal mapped its findings to the OWASP Top 10:2025 categories. It found two issues, both fixed and confirmed by retest. Checks that found nothing apply to the build tested."
How do you ensure one customer cannot access another customer's data?Is our data kept apart from other customers' data in a shared system?Findings and negative results from tests run between two customer organisations and between roles, recorded for each account and record tried.Covers the functional areas tested on that build. New features need their own check."Access between customer organisations was tested in March across accounts, contacts and exports. One issue was found and fixed, confirmed by retest in May. New features are added to the next test's scope."
Is customer data encrypted in transit and at rest?Could our data be read on the network, or from a stolen disk or backup?In transit: the certificate and transport security results recorded for each public hostname in an external check. At rest: your hosting and database provider's configuration and documentation.A web test sees transport security from outside. It cannot see how stored data is encrypted."Data in transit uses TLS on every public hostname, as recorded in our March external check. Stored data is encrypted by our managed database provider, whose documentation is attached."
Are you ISO 27001 certified, or can you provide a SOC 2 report?Has an auditor examined how you manage security?Only the certificate or the SOC 2 report itself.A penetration test report is neither a certification nor an audit report, and does not substitute for one."We are not ISO 27001 certified and do not have a SOC 2 report. Our security controls are summarised in the attached overview, and our customer portal was penetration tested in March 2026."
Are you PCI DSS compliant?If you take card payments, has your payment setup been checked against the card industry's standard, and by whom?Your own PCI DSS validation documents, and any scan reports from an Approved Scanning Vendor or assessment by a Qualified Security Assessor.A web application test is not an Approved Scanning Vendor scan or a Qualified Security Assessor assessment."Card payments are taken on our payment provider's hosted page, and card numbers do not pass through our servers. Our PCI DSS validation documents are attached." Use the first sentence only if your developer confirms it.
Was the test carried out by a CREST-accredited provider?Does the tester meet our procurement policy?The provider's name and the accreditation it holds, if any.If the policy requires CREST, only a CREST-accredited provider meets it. A report from a provider without it does not."Our March test was carried out by a provider that is not CREST accredited. If your policy requires CREST, please tell us and we will engage an accredited provider for the next test."
How was our data handled during testing, and was production data used?Were real customer records exposed during the test, and where is the evidence now?The environment named in the report's scope, the list of test data left behind, evidence shown with secrets redacted, and the data-handling terms in the provider's statement of work.The report shows what was left behind and what was redacted. Where raw evidence is stored, and when it is deleted, is set by the agreement, so ask for it in writing."Testing ran on a staging copy using test accounts and seeded data. The report listed two test records left behind, which we removed in April. Credentials are redacted in the report's evidence."
Can you provide an attestation or summary letter from your tester?Can we see proof of the test without reading the full report?The executive summary and scope pages from the report. A short letter from the provider stating what was tested and when, where one was agreed at scoping.A letter describes the test. It adds no assurance the report does not contain, and whether a provider issues one is worth asking before the test starts."We can share the executive summary and the scope as tested under a confidentiality agreement. A summary letter was not part of our engagement with the tester."
Do you perform secure code review or static analysis before release?Are weaknesses caught before code reaches production?The report's scope, which states whether source code was reviewed, and your own development records, such as pull request reviews or dependency alerts.A test without code access does not evidence code review."Our March penetration test did not include source code review. Each change is reviewed by a second developer before release, and dependency alerts are checked weekly."
Is multi-factor authentication enforced for all users?Can a stolen password alone open an account that matters?For the application: what the login and session tests observed for each role. For staff email, cloud and admin consoles: your identity provider's settings."All users" usually hides two answers, staff and customers. A test of the application says nothing about staff email."Multi-factor authentication is enforced for all staff accounts and for administrators in our application. It is optional for customer accounts, which can only see their own records."
How do you identify and fix vulnerabilities between penetration tests?What happens in the months the last report does not cover?Your patching and update records, dependency alerts, any scheduled scan results, and the remediation plan from the last report showing what was done.A penetration test is a point-in-time event. This question is about your process, and the report shows only its starting point."Dependencies and plugins are updated monthly, and security releases are applied within a week. The remediation plan from our March test is complete apart from one item scheduled for July."
Have you had a security incident or data breach in the last three years?Is there a history that changes the risk, and would you tell us about the next one?Your incident register and your incident response plan.A test cannot evidence breach history, or how you would respond to an incident."We have had no incidents involving customer data in the last three years. Our incident response plan names who decides on notification and how affected customers are told." Write this only if your records support it.

Typical question

What the reviewer is really asking

  1. Do you conduct annual penetration testing of your application?
    What the reviewer is really asking
    Is an outside party testing the system we will use, and is that a repeating programme or a one-off?
    Evidence that answers it
    The report's scope as tested: targets, roles, build and dates. A booking or statement of work for the next test, if one exists.
    Where that evidence stops
    One report shows one test. It does not show a programme, or anything outside its scope.
    Example answer that does not overstate
    "Our customer portal and its API were penetration tested in March 2026 by an external provider. This was our first test. The next is planned for March 2027."
  2. When was your last penetration test, and can you share the results?
    What the reviewer is really asking
    Is the evidence recent enough to describe the system as it runs today?
    Evidence that answers it
    The report date and version, the executive summary and the findings table.
    Where that evidence stops
    Releases since the test are not covered. A large change since then, such as a new module, sits outside the evidence.
    Example answer that does not overstate
    "Last tested in March 2026, before our billing module launched in June. We can share the executive summary and findings table under a confidentiality agreement. Billing is in scope for the next test."
  3. Were any critical or high-risk findings identified, and have they been remediated?
    What the reviewer is really asking
    Is a known serious weakness still open in the system we will rely on?
    Evidence that answers it
    The findings table with business-risk ratings and CVSS v3.1 scores, the remediation plan, and the retest section marking each finding fixed, partially fixed, still reproducible or not retested.
    Where that evidence stops
    Fixed means fixed in the retested build. A finding marked not retested is unverified, whatever the developer reports.
    Example answer that does not overstate
    "The test reported one critical and two high findings. The May retest recorded the critical finding and one high finding as fixed. The other is partially fixed; the remaining change ships in July and will be retested then."
  4. Was the test performed by an independent third party?
    What the reviewer is really asking
    Could the tester have been marking their own work?
    Evidence that answers it
    The provider's written statement of whether it built, hosts or maintains the system. A Clever Ops report records who built and who hosts a system when that is Clever Ops.
    Where that evidence stops
    A review by the builder or host is useful for fixing problems but is not independent evidence.
    Example answer that does not overstate
    "The March review was carried out by the agency that built and hosts the application, and its report says so. It was not an independent test. An independent test by a separate provider is booked for August."
  5. What methodology or standard was the test performed against?
    What the reviewer is really asking
    Was the testing structured, or was someone clicking around the site?
    Evidence that answers it
    The report's methodology section naming its references, such as the OWASP Web Security Testing Guide and ASVS 5.0, and the OWASP Top 10:2025 category on each finding.
    Where that evidence stops
    Referencing a framework describes how the work was organised. It is not a certification against that framework.
    Example answer that does not overstate
    "Testing referenced the OWASP Web Security Testing Guide and ASVS 5.0, and each finding is mapped to an OWASP Top 10:2025 category. The test was not a certification against those documents."
  6. Is your application protected against the OWASP Top 10?
    What the reviewer is really asking
    Has anyone checked for the common categories of web application risk?
    Evidence that answers it
    Findings mapped to OWASP Top 10:2025 categories, plus the negative results for checks that found nothing.
    Where that evidence stops
    The OWASP Top 10:2025 is an awareness document on the most critical security risks to web applications, not a pass mark. A test cannot show that no issue in those categories exists.
    Example answer that does not overstate
    "Our March test of the customer portal mapped its findings to the OWASP Top 10:2025 categories. It found two issues, both fixed and confirmed by retest. Checks that found nothing apply to the build tested."
  7. How do you ensure one customer cannot access another customer's data?
    What the reviewer is really asking
    Is our data kept apart from other customers' data in a shared system?
    Evidence that answers it
    Findings and negative results from tests run between two customer organisations and between roles, recorded for each account and record tried.
    Where that evidence stops
    Covers the functional areas tested on that build. New features need their own check.
    Example answer that does not overstate
    "Access between customer organisations was tested in March across accounts, contacts and exports. One issue was found and fixed, confirmed by retest in May. New features are added to the next test's scope."
  8. Is customer data encrypted in transit and at rest?
    What the reviewer is really asking
    Could our data be read on the network, or from a stolen disk or backup?
    Evidence that answers it
    In transit: the certificate and transport security results recorded for each public hostname in an external check. At rest: your hosting and database provider's configuration and documentation.
    Where that evidence stops
    A web test sees transport security from outside. It cannot see how stored data is encrypted.
    Example answer that does not overstate
    "Data in transit uses TLS on every public hostname, as recorded in our March external check. Stored data is encrypted by our managed database provider, whose documentation is attached."
  9. Are you ISO 27001 certified, or can you provide a SOC 2 report?
    What the reviewer is really asking
    Has an auditor examined how you manage security?
    Evidence that answers it
    Only the certificate or the SOC 2 report itself.
    Where that evidence stops
    A penetration test report is neither a certification nor an audit report, and does not substitute for one.
    Example answer that does not overstate
    "We are not ISO 27001 certified and do not have a SOC 2 report. Our security controls are summarised in the attached overview, and our customer portal was penetration tested in March 2026."
  10. Are you PCI DSS compliant?
    What the reviewer is really asking
    If you take card payments, has your payment setup been checked against the card industry's standard, and by whom?
    Evidence that answers it
    Your own PCI DSS validation documents, and any scan reports from an Approved Scanning Vendor or assessment by a Qualified Security Assessor.
    Where that evidence stops
    A web application test is not an Approved Scanning Vendor scan or a Qualified Security Assessor assessment.
    Example answer that does not overstate
    "Card payments are taken on our payment provider's hosted page, and card numbers do not pass through our servers. Our PCI DSS validation documents are attached." Use the first sentence only if your developer confirms it.
  11. Was the test carried out by a CREST-accredited provider?
    What the reviewer is really asking
    Does the tester meet our procurement policy?
    Evidence that answers it
    The provider's name and the accreditation it holds, if any.
    Where that evidence stops
    If the policy requires CREST, only a CREST-accredited provider meets it. A report from a provider without it does not.
    Example answer that does not overstate
    "Our March test was carried out by a provider that is not CREST accredited. If your policy requires CREST, please tell us and we will engage an accredited provider for the next test."
  12. How was our data handled during testing, and was production data used?
    What the reviewer is really asking
    Were real customer records exposed during the test, and where is the evidence now?
    Evidence that answers it
    The environment named in the report's scope, the list of test data left behind, evidence shown with secrets redacted, and the data-handling terms in the provider's statement of work.
    Where that evidence stops
    The report shows what was left behind and what was redacted. Where raw evidence is stored, and when it is deleted, is set by the agreement, so ask for it in writing.
    Example answer that does not overstate
    "Testing ran on a staging copy using test accounts and seeded data. The report listed two test records left behind, which we removed in April. Credentials are redacted in the report's evidence."
  13. Can you provide an attestation or summary letter from your tester?
    What the reviewer is really asking
    Can we see proof of the test without reading the full report?
    Evidence that answers it
    The executive summary and scope pages from the report. A short letter from the provider stating what was tested and when, where one was agreed at scoping.
    Where that evidence stops
    A letter describes the test. It adds no assurance the report does not contain, and whether a provider issues one is worth asking before the test starts.
    Example answer that does not overstate
    "We can share the executive summary and the scope as tested under a confidentiality agreement. A summary letter was not part of our engagement with the tester."
  14. Do you perform secure code review or static analysis before release?
    What the reviewer is really asking
    Are weaknesses caught before code reaches production?
    Evidence that answers it
    The report's scope, which states whether source code was reviewed, and your own development records, such as pull request reviews or dependency alerts.
    Where that evidence stops
    A test without code access does not evidence code review.
    Example answer that does not overstate
    "Our March penetration test did not include source code review. Each change is reviewed by a second developer before release, and dependency alerts are checked weekly."
  15. Is multi-factor authentication enforced for all users?
    What the reviewer is really asking
    Can a stolen password alone open an account that matters?
    Evidence that answers it
    For the application: what the login and session tests observed for each role. For staff email, cloud and admin consoles: your identity provider's settings.
    Where that evidence stops
    "All users" usually hides two answers, staff and customers. A test of the application says nothing about staff email.
    Example answer that does not overstate
    "Multi-factor authentication is enforced for all staff accounts and for administrators in our application. It is optional for customer accounts, which can only see their own records."
  16. How do you identify and fix vulnerabilities between penetration tests?
    What the reviewer is really asking
    What happens in the months the last report does not cover?
    Evidence that answers it
    Your patching and update records, dependency alerts, any scheduled scan results, and the remediation plan from the last report showing what was done.
    Where that evidence stops
    A penetration test is a point-in-time event. This question is about your process, and the report shows only its starting point.
    Example answer that does not overstate
    "Dependencies and plugins are updated monthly, and security releases are applied within a week. The remediation plan from our March test is complete apart from one item scheduled for July."
  17. Have you had a security incident or data breach in the last three years?
    What the reviewer is really asking
    Is there a history that changes the risk, and would you tell us about the next one?
    Evidence that answers it
    Your incident register and your incident response plan.
    Where that evidence stops
    A test cannot evidence breach history, or how you would respond to an incident.
    Example answer that does not overstate
    "We have had no incidents involving customer data in the last three years. Our incident response plan names who decides on notification and how affected customers are told." Write this only if your records support it.

What a security questionnaire asks you to prove

A security questionnaire is how a customer's security, risk or procurement team decides whether to trust a supplier with its data, its systems or its name.

It tends to arrive late in a sale or at renewal, as a spreadsheet, a form inside a supplier portal or a document with a due date. You may see it called a vendor security questionnaire, a supplier security assessment or a cyber security questionnaire.

Whatever the label, the reviewer wants three things from each answer: a statement they can rely on, the evidence behind it, and a clear line around what it covers.

Some customers write their own questions. Others start from a published set. The Cloud Security Alliance publishes CAIQ v4.1, a set of Yes/No questions a cloud consumer and cloud auditor may wish to ask of a cloud provider, to ascertain its compliance with the Cloud Controls Matrix (CCM).

Suppliers of digital products to Australian schools may come across Safer Technologies 4 Schools (ST4S), a standardised approach to evaluating digital products and services used by schools against a nationally consistent security and privacy control framework.

A customer's own spreadsheet can borrow wording from sets like these, so similar questions turn up from very different customers.

Treat every answer as a statement you may later be asked to support. A questionnaire can be attached to a contract or reopened at renewal, and a yes that turns out to cover less than it implied is harder to walk back than a qualified answer given at the start.

Sort the questions before you answer any of them

A questionnaire usually mixes three kinds of question, and only one kind is mainly about testing.

Sorting them first stops two mistakes: hunting for a test to answer a policy question, and answering a testing question with a policy.

  • Organisation questions cover policies, staff screening and training, who is responsible for security, insurance, incident handling, breach history and how you choose your own suppliers. No security test answers these. They come from your documents and records.
  • Control questions ask whether a control exists: multi-factor authentication, encryption, log retention, backups and restores, removing access when staff leave. A test can speak to the few that are visible from a browser or an API on the application it covered, such as transport encryption or how sessions end. The rest are configuration facts about your systems and providers.
  • Assurance questions ask how you know: when the system was last tested, by whom, against what, what was found, whether it was fixed, and which certifications or audit reports you hold. This is where a penetration test or security review does most of its work, and where an answer can easily say more than the test showed.

What a penetration test report can evidence

A test report is dated evidence about one system, tested in one window, against an agreed scope.

Read that way, its sections line up closely with the assurance questions. Whether that evidence is enough is for the customer or auditor reading it to decide; the report's job is to make the decision easy to reach. Our sample report, built from a lab application, shows the format.

  • Scope as tested: the targets, roles, build and dates tested, and what was out of scope or not performed. This answers "when was it last tested" for those targets, and the exclusions list keeps an answer from reaching further than the test did.
  • Method: the frameworks the testing referenced. Clever Ops reports reference the OWASP Web Security Testing Guide and the Application Security Verification Standard 5.0, and map each finding to its OWASP Top 10:2025 category. That tells a reviewer how the work was structured. It does not say the application meets any of those documents.
  • Findings and severity: each rated finding has a CVSS v3.1 technical score on the 0.0 to 10.0 scale and a separate business-risk rating, alongside its evidence, impact and fix. A reviewer asking about critical or high findings can read both ratings side by side.
  • Negative results: checks that found nothing are recorded for the build and accounts tested. They support a narrow statement, such as "access between customer organisations was tested in these areas and no issue was found", and nothing wider.
  • Remediation and retest: the remediation plan orders the fixes, and after the retest each finding is marked fixed, partially fixed, still reproducible or not retested. This is the evidence behind "were the issues fixed?", and a finding marked not retested cannot honestly be answered as fixed.
  • Test data and redaction: the report lists any test data left behind for your team to remove, and shows evidence with secrets redacted. Both help with questions about how the tester treated your data.
  • Who built and who hosts the system: when Clever Ops built or hosts it, the work is a security review rather than an independent test, and the report records who built and who hosts it.

Questions a test report cannot answer for you

A report is strong evidence inside its scope and silent outside it.

Stretching it over the questions below invites a follow-up you cannot answer, and puts the answers it does support in doubt.

  • Certification and audit questions. Only the certificate or audit report itself answers a question such as "Are you certified?" or "Do you have a SOC 2 report?" Clever Ops is not ISO 27001 certified, holds no SOC 2 report and issues neither, so a test from us cannot stand in for one.
  • Payment card evidence. An Approved Scanning Vendor conducts external vulnerability scanning to validate adherence with the external scanning requirements of PCI DSS Requirement 11.3.2, and Qualified Security Assessor companies are qualified by the PCI Security Standards Council to validate an entity's adherence to PCI DSS. Clever Ops is not an Approved Scanning Vendor or a Qualified Security Assessor, so that evidence comes from a provider that is.
  • Office network, staff devices, email and cloud accounts. A web application test covers what was in its scope. Questions about laptops, patching of staff machines or phishing training are answered from your IT provider's records, or by a network specialist's test.
  • Backups, logging and monitoring. A test may note logging gaps it could observe, but whether backups restore, how long logs are kept and who watches the alerts are facts about how you operate.
  • Incident response and breach history. Answer from your incident response plan and your incident records. Some questionnaires ask about notifiable breaches: under the Notifiable Data Breaches scheme, an organisation or agency covered by the Privacy Act 1988 must notify affected individuals and the OAIC when a data breach is likely to result in serious harm to a person whose personal information is involved.
  • Tester accreditation. If the customer's policy requires a CREST-accredited tester, only a CREST-accredited provider meets it. CREST awards accreditation to organisations after an application, evidence submission and review, and its accreditation disciplines include penetration testing. Clever Ops is not CREST accredited, so we refer that work to a provider that is.

Penetration testing for SOC 2 and ISO 27001 audits

If you are working towards SOC 2 or ISO 27001, the same test report may be read twice: once by a customer deciding whether to trust the system they will use, and once by your auditor, who decides what, if anything, it counts as evidence of. Neither reader is bound by the other's view, so check what each one needs.

According to the AICPA, System and Organization Controls (SOC) is a suite of service offerings CPAs may provide on system-level controls of a service organisation or entity-level controls of other organisations, and the AICPA publishes guidance on SOC 2 reporting on an examination of controls at a service organisation relevant to security, availability, processing integrity, confidentiality or privacy.

A penetration test report is not that examination. Whether it counts as evidence for any control in your SOC 2 work is your auditor's call.

ISO 27001 penetration testing follows the same logic: your auditor decides what evidence they need, and a test report is one thing you can offer them. Clever Ops is not ISO 27001 certified and does not certify anyone. A report from us records what was tested, what was found and what was fixed, for your auditor to weigh.

Before scoping a test for an audit, put these four questions to the auditor and bring the answers to the scoping call.

  • Which systems do you expect to see tested, and does that include the application our customers use?
  • How recent does the test need to be at the time of the audit?
  • Do you expect the tester to be independent of the people who built and run the system?
  • Do you want evidence that findings were retested, or is the remediation plan enough?

When the builder is also the tester

Independence is where a build or hosting relationship matters. If an auditor or customer asks for an independent test, the provider should have had no part in building or running the system. When Clever Ops built or hosts a system, we call the work a security review, and the report says who built and who hosts it, so it is not presented as an independent test. It is still useful before a launch or a major change. It simply answers a different question.

How to answer when the honest answer is "not yet"

Sooner or later a questionnaire asks something you cannot answer with a clean yes.

A blank, a vague yes, or "in progress" with no date all tend to draw follow-up questions. A qualified answer is quicker to accept because it already contains the follow-up. It has four parts.

  • What is true now, with its scope and date: "Our customer portal and its API were penetration tested in March 2026."
  • Where that stops: "The test did not include our mobile app or our office network."
  • What you are doing about the gap, with a date you can keep: "A test of the mobile app is booked for August 2026."
  • Any control that covers the gap in the meantime, only if it is real: "The mobile app calls the same API, which was in scope."

Answer the question that was asked

If the question is "Do you conduct annual penetration testing?" and you have had one test, say you have had one test, when it was, and whether the next is booked. "Yes" claims a programme, not a single event. In the same way, "Are all findings remediated?" needs the retest status of each finding, not the developer's view that the work is done.

Say open findings out loud

An open finding with a plan is a normal answer: "One high-risk finding is partially fixed. The remaining change ships in July and will be retested then." The answer that damages trust is one that says nothing is open, sitting beside a report whose status column says otherwise.

Keep one answer register

Save each completed questionnaire with the evidence you cited for each answer. When the next one arrives, start from the last set and update the dates. Answers written from scratch each time drift apart, and a customer at renewal, or two customers comparing notes, can spot the contradiction.

Sharing test evidence without oversharing

A full report contains reproduction steps for every finding, including any not yet fixed.

Before it leaves the business, decide what the reviewer needs to see, and send that.

  • The executive summary and the scope as tested answer when, what and by whom.
  • The findings table with ratings and retest statuses answers what was found and whether it was fixed, without the step-by-step evidence.
  • The full report, if the reviewer needs it, goes under a confidentiality agreement and preferably after the retest, so the status column is current.

If the customer asks for a letter

Some customers ask for a short letter instead of the report, stating what was tested and when. If yours might, raise it on the scoping call. Whether a letter can be provided, and what it says, is agreed with the provider at that point rather than assumed at the end. A letter describes a test. It adds no assurance that the report itself does not hold.

Match your wording to the report

Whatever you send, keep the dates and the names of systems exactly as the report writes them. A reviewer comparing an answer with its evidence notices when "the platform" in the answer was "the customer portal" in the report, and starts checking everything else more closely.

Building an evidence pack before the next questionnaire

If questionnaires are becoming part of how you win work, assemble the evidence once and keep it current.

Reviewers ask for the same documents in different words, and a pack you can attach turns chasing into matching.

  • The latest test summary, scope as tested and retest statuses, with the full report held back for requests under a confidentiality agreement.
  • A one-page overview of your security controls, written in the same terms as your questionnaire answers.
  • Your hosting, database and payment providers' own security documentation, for the controls they run on your behalf.
  • Your incident response plan, or a summary of it, and the name of the person who owns it.
  • A dated plan for the next test, timed so the test, the fixes and the retest all finish before your busiest renewal period rather than during it.
  • Your answer register, so each new questionnaire starts from answers that have already been checked.

Frequently asked questions

A set of questions a customer sends a supplier before trusting it with data, system access or a contract. It covers how the supplier is organised, which security controls it runs and what assurance sits behind them, such as testing and audits. Some customers write their own; others start from a published set such as the Cloud Security Alliance's CAIQ.

Not for most of it. Policy, staffing, backup and incident questions are answered from your own records. A test matters for the assurance questions: when the system was last tested, against what, and whether the findings were fixed. If the questionnaire asks for a recent test and you have not had one, say so, give the date one is booked, and scope it to the system that customer will use.

Ask your auditor. They decide what evidence your audit needs, so their answer should shape the scope and timing of any test. What a test report offers them is dated evidence of what was tested, what was found and what was fixed. Clever Ops does not issue SOC 2 reports or ISO 27001 certification, so a report from us is evidence for your auditor to weigh, not a certificate.

Start smaller. The executive summary, the scope as tested and the findings table with retest statuses answer the usual assurance questions without exposing reproduction steps for anything still open. If the reviewer needs the full report, send it under a confidentiality agreement, after the retest where you can.

It can review its own work, but the result is not independent evidence. When Clever Ops built or hosts a system, the work is called a security review and the report states who built and who hosts it. If the questionnaire or the customer asks for an independent tester, choose a provider with no part in building or running the system.

Answer with what is true today and the date the test is booked, and offer to send the summary once the retest is done. A clear plan with a date is easier for a reviewer to accept than a rushed test whose findings are still open when you reply.

No. A report shows what was found in the scope, build and window tested. It does not cover policies, staff, backups or systems outside that scope, and a check that found nothing applies only to what was checked. Answer each question from the evidence that covers it.

Scope the right test for your application

Tell us what you have built, who uses it and when it needs to launch. You will get a plain-English recommendation: a website audit, a web application test or a referral to a specialist.