Skip to main content
Security templateReviewed

Vendor security questionnaire: forty questions to send a software vendor, and how to judge the answers

For the business doing the buying.

Each question names the evidence to ask for, the person on your side who judges the answer, what a weak answer looks like and what to ask next, with a review date to fill in. The page also shows how to cut the forty down for a low-risk vendor, and why a completed questionnaire is a claim rather than proof.

Forty-question vendor security questionnaire with evidence and red flags

Copy the table into a spreadsheet, one row per question. Send the vendor the question and evidence columns for the rows its tier needs, and keep the owner, red flag and review columns for your own team.

How to use it

References run G1 to G6 governance, T1 to T7 application testing, H1 to H6 data hosting, S1 to S5 sub-processors, A1 to A6 access control, I1 to I5 incidents and R1 to R5 resilience. Replace each prompt in round brackets with a date when you check the answer.

40 rows

Forty-question vendor security questionnaire with evidence and red flags

QuestionEvidence to requestAnswer owner (your side)Red flag and next questionReview date
G1. Who at your company is accountable for security, and who stands in for them when they are away?The names and roles of the accountable person and a deputy, taken from the security policy or an organisation chart.Contract owner"Our IT team" or "everyone" with no named person, or one founder who is also the only developer. Ask for the name and phone number your team would call if the product were breached at night.Checked (date). Next: at renewal, or when either person leaves.
G2. Which written security policies apply to the product, and when was each one last reviewed and approved?A list of policy titles with version numbers, owners and approval dates. The full documents under a confidentiality agreement, if the vendor's tier calls for them.IT leadUndated policies, a template with another company's name left in, or policies about an office network that say nothing about the product. Ask for the approval record of the current versions.Checked (date). Next: at renewal.
G3. Which independent certifications or audit reports cover this product, who issued them, and what scope and dates do they cover?The document itself: a SOC 2 report, an ISO/IEC 27001 certificate or a CSA STAR Level 2 listing, showing the issuer, the dates and the service it covers.IT leadA badge on the website instead of the document, or a certificate that covers a parent company, an office or a different service. The NCSC asks whether a certification's scope covers the parts of the service you will be using. Ask for the scope statement and the issue and expiry dates.Checked (date). Next: (certificate expiry, or the end of the report period).
G4. How can outsiders report a vulnerability in the product, and what happens to a report once it arrives?The published vulnerability disclosure policy, the security.txt file at /.well-known/security.txt if there is one, and the triage procedure with its response times.Technical leadNo reporting route at all, or a contact address that bounces. Check the policy page and the file yourself. Ask when the last report arrived and how long it took to resolve.Checked (date). Next: when the Expires date in the security.txt file passes, or at renewal.
G5. How are staff and contractors who can reach customer data trained in security, and how is completion recorded?Training records showing role-specific content and completion dates for everyone with production or support access, contractors included.IT leadOne generic video a year with no completion record, or contractors left out. MVSP asks for role-specific security training relevant to each person's business function. Ask for the completion rate among staff who can reach production.Checked (date). Next: at renewal.
G6. Which of the answers in this questionnaire will you commit to in the contract?The vendor's security schedule or data processing terms, marked against the rows of this questionnaire they cover.Contract ownerGenerous answers on the form that appear nowhere in the contract, or "our terms are fixed" for a full-tier service. Ask which rows they will put in writing, and treat the rest as unconfirmed.Checked (date). Next: at each contract change or renewal.
T1. When was the product last penetration tested, and what did the scope include?The executive summary and the scope as tested from the latest report: the application, the API, the roles, the environment, the dates and the exclusions.Technical leadA test of the marketing website, a scope that leaves out the API or the roles you will use, or a test date before the last major release. Ask for the scope page and the release history since the test.Checked (date). Next: twelve months after the report date, or after the vendor's next major release.
T2. Who carried out that test, and were they independent of the people who built and run the product?The testing provider's name, and a statement of whether it built, hosts or maintains the product.Technical leadThe development agency reviewing its own work, or an internal team exercise described as independent. Ask how the tester relates to the builder. If your policy requires a CREST-accredited tester, ask for the provider's name and confirm the accreditation before you rely on the report.Checked (date). Next: with T1.
T3. Which critical and high-risk findings were reported, and what is the retest status of each one now?The findings table, with a severity for each finding and its retest status: fixed, partially fixed, still reproducible or not retested.Technical lead"All findings remediated" with no retest behind it, or a count of findings with no statuses. A finding marked not retested has not been shown to be fixed. Ask for the retest date and the build it ran on.Checked (date). Next: (date the last open finding is due for retest).
T4. Which methods or standards did the testing reference, and how are the findings mapped to them?The methodology section of the report, naming references such as the OWASP Web Security Testing Guide or ASVS 5.0, with each finding mapped to a category.Technical lead"Tested against the OWASP Top 10" offered as a pass mark. Naming a document describes how the testing was organised, not what it showed. Ask which checks were run for the roles you will use, and what the checks that found nothing covered.Checked (date). Next: with T1.
T5. How do you find and fix vulnerabilities in your code and third-party components between tests, and how fast?The patching timeframes by severity from the vendor's policy, and last quarter's dependency alerts with the date each one was closed.Technical lead"We run automated scans" with no timeframes or records. MVSP's time-to-fix control asks for patches to application vulnerabilities that materially impact security within 90 days of discovery. Ask for the vendor's own timeframes and its record against them.Checked (date). Next: at renewal.
T6. Can you provide a software bill of materials (SBOM), or another current list of the product's third-party components?An SBOM in a standard, machine-readable format, which CISA's Secure by Demand guide suggests asking for, or a dated list of components with their versions.Technical leadNo inventory at all. OWASP Top 10:2025 says you are likely vulnerable to supply chain failures if you do not carefully track the versions of all the components you use, including nested dependencies. Ask how the vendor would learn that a component in the product had a published vulnerability.Checked (date). Next: at renewal.
T7. Will you allow security testing of the product commissioned by a customer, and on what terms?The contract clause or published testing terms, naming the environment, the notice required and anything excluded.Technical leadA flat refusal from a vendor that has no recent independent test to offer instead. MVSP's customer testing control asks vendors to let customers or their delegates test on request, on production or on a non-production environment that closely resembles it in functionality. Ask for written terms before anyone sends a single request.Checked (date). Next: before any test is commissioned.
H1. In which countries is customer data stored, processed and backed up, by which hosting providers, and from where do support and development staff reach it?A list of hosting providers and regions for production data and backups, and the countries that support and development staff work from.Privacy lead"In the cloud" with no region, or storage in Australia with backups or support handled elsewhere. The NCSC asks whether a supplier offshores any part of its service, including support, development or maintenance, and whether it will tell you if locations change. Ask for notice before any location changes.Checked (date). Next: at renewal, or when the vendor announces a new region or hosting provider.
H2. How is customer data encrypted in transit and at rest, including backups, and who manages the keys?The vendor's security overview or architecture note naming the encryption used for connections, databases, file storage and backups, and who holds the keys.IT lead"Industry-standard encryption" with no detail, or encryption at rest that stops short of backups and file storage. Ask which data stores the answer covers, one by one.Checked (date). Next: at renewal.
H3. Which types of data will the product hold for you, and do you keep a current diagram of how that data flows?A list of the data types the product processes, and a data flow diagram showing where each one enters and where it is stored.Privacy leadThe vendor cannot say what it holds or where it ends up. MVSP lists both a list of sensitive data types and an up-to-date data flow diagram among its controls. Ask for the diagram covering the features you will switch on.Checked (date). Next: when you start using a new feature or sending a new type of data.
H4. How is one customer's data kept apart from another customer's inside the product, and was that separation tested?The architecture description of how customer accounts are separated, and the findings or negative results from test checks run between two customer accounts.Technical lead"Every customer has their own login" offered as separation. Ask whether the last test used accounts in two customer organisations, and what it found.Checked (date). Next: with T1.
H5. Can we export all of our data in a usable format, at any time and at contract end?The export documentation, a sample export, and the contract clause on returning data.Operations leadExport only through a support request, only for some records, or in a format nothing else can read. Run an export yourself during the trial and open the file, then ask how a full export of a large account is delivered and how long it takes.Checked (date). Next: run a test export once a year.
H6. Is production data ever copied into test, development or support environments?The policy on data in non-production environments, and a description of how test data is created or masked.Privacy leadDevelopers working on copies of live customer records. MVSP, in its customer testing control, says non-production environments should not contain production data. Ask how test data is made and who can reach any copies.Checked (date). Next: at renewal.
S1. Which third parties can access customer data, what does each one do, and where is it based?The current sub-processor list, naming each company, its role and its location.Privacy leadCategories such as "cloud provider" instead of company names, or no list at all. MVSP asks vendors to keep a list of third-party companies with access to customer data and make it available to clients on request. Ask for the named companies.Checked (date). Next: at renewal, and on each change notice.
S2. How will you tell us before adding or replacing a sub-processor, and can we object?The notification clause, the notice period and the channel used, such as an email list or a web page with a change history.Contract ownerChanges made quietly on a web page with no notice. The NCSC asks whether you would wish to be informed if a supplier changes subcontractors. Ask for notice before the change takes effect, and a route to object or to leave.Checked (date). Next: on each change notice.
S3. How do you assess the security of your own suppliers, and when did you last review your main hosting provider?The vendor's supplier review procedure, and the date and outcome of its last review of its two most important sub-processors.IT lead"Our hosting provider handles security" with no review record. MVSP asks vendors to assess third parties with access to customer data every year, against the latest MVSP release. Ask for the last review dates.Checked (date). Next: at renewal.
S4. Which of your commitments in this contract are passed down to the sub-processors that hold our data?The flow-down clauses in the vendor's sub-processor agreements, or a summary of them.Contract ownerCommitments on incident notice or data location that the vendor's own suppliers are not bound by. The NCSC asks whether controls in your contract with a supplier must be passed down to all subcontractors. Ask which commitments flow down, row by row.Checked (date). Next: at renewal.
S5. Which AI services or other outside processors receive our data through features of the product, and can that data be used to train models?A list of features that send data to an outside AI or processing service, the provider behind each one, the data it receives, and the setting or contract term on model training.Privacy leadAI features switched on by default with no list of providers, or a general line about not selling data in place of an answer on training. Ask whether each feature can be switched off for your account, and how.Checked (date). Next: whenever the vendor announces a new AI feature.
A1. Do you support single sign-on with our identity provider and multi-factor sign-in for every user, and on which plans?The product documentation for single sign-on and multi-factor sign-in, the plans that include each, and the settings as seen in a trial account.IT leadSingle sign-on only on the top plan, multi-factor sign-in that is optional and off by default, or a shared default password at setup. CISA's Secure by Demand guide asks whether a manufacturer supports standards-based single sign-on at no additional cost and, where it manages authentication, enables multi-factor authentication by default at no cost. Ask whether administrators can enforce it for every user.Checked (date). Next: at renewal, and whenever you change plan.
A2. Which of your staff can access customer data, how is that access approved, and when was it last reviewed?The access policy, the number of staff with access to production data, and the date and outcome of the last access review.IT lead"Only authorised staff" with no number and no review record, or every support agent able to open every customer's account. MVSP asks vendors to limit sensitive data access to users with a legitimate need and to review access regularly. Ask whether staff access to your data is logged, and whether you can see that log.Checked (date). Next: at renewal.
A3. Is multi-factor authentication required for staff who reach production systems or customer data remotely?The identity provider's enforcement setting and the list of systems it covers: hosting console, database, code repository and support tools.IT leadMulti-factor sign-in on email only, with the hosting console or support tools left on passwords. MVSP asks that remote access to customer data or production systems require multi-factor authentication. Ask for the list of systems.Checked (date). Next: at renewal.
A4. How quickly is access removed when a staff member or contractor leaves?The leaver procedure with its timeframe, and a recent leaver record with the name removed, showing the date access ended in each system.IT leadNo timeframe, or contractors handled outside the process. MVSP asks for redundant accounts and expired access grants to be deactivated in a timely manner. Ask for the most recent leaver record.Checked (date). Next: at renewal.
A5. What does the product log about sign-ins, administrator actions and data access, how long are those logs kept, and can we export them?The audit log documentation, the retention period, the plans that include it, and a sample log screen from a trial account.IT leadAudit logs only on the top plan, or kept for a few days. CISA's Secure by Demand guide says cloud and SaaS providers should retain security logs and make them available to customers for at least six months at no additional charge, and MVSP sets 30 days at no additional charge as its minimum. Ask whether the vendor's own staff access to your data appears in the log.Checked (date). Next: at renewal.
A6. How does the product control what each role inside our account can see and change?The role and permission table from the documentation, checked in a trial account with two different roles.Technical leadOnly an administrator role and a user role, so anyone who needs one setting gets all of them. Ask whether permissions can be limited by team, by record type or by action.Checked (date). Next: when you add a team or a new kind of user.
I1. How quickly will you notify us of a security incident affecting our data, through which channel, and what will the notice contain?The contract clause with the notice period in hours or days, the channel and the named contacts, and the notice template if there is one.Contract owner"Without undue delay" with no number, or notice only through a public status page. MVSP's incident handling control asks vendors to notify relevant parties of a breach affecting sensitive information no later than 72 hours after discovery, with its nature, contacts, consequences and the measures taken. Ask for a number in the contract and a named contact on each side.Checked (date). Next: at renewal, and after any incident.
I2. Have you had a security incident or breach affecting customer data in the last three years, and what changed afterwards?A summary of each incident: the date, what happened, the data affected, which customers were told, and the changes made since.Contract owner"None" from a vendor whose own announcements say otherwise, or an incident followed by no change. Search the vendor's name with "breach" and "incident" before you accept a none. Ask what changed and how they checked that it worked.Checked (date). Next: at renewal.
I3. Do you have a written incident response plan, and when was it last exercised?The plan's contents page and its owner, and the date and lessons of the last exercise or real incident.IT leadA plan with no step for telling customers, or one that has never been exercised. Ask what the last exercise found and what was changed because of it.Checked (date). Next: at renewal.
I4. Who will our team work with during an incident, and how will you include us in the response?Named incident contacts with phone numbers, and the vendor's process for working with customers while an incident is under way.IT leadA general support inbox as the only contact. Ask for a phone number that is answered out of hours, and add it to your own incident contact sheet.Checked (date). Next: every six months, by checking the number still answers.
I5. After an incident, what will you share with us: logs, a timeline, a root cause report?The contract or policy term on post-incident reports, and on keeping logs long enough for an investigation.IT leadNo commitment beyond a status page update, or logs deleted on a short cycle. Ask how long logs are kept (row A5) and whether affected customers receive a written root cause report.Checked (date). Next: at renewal.
R1. How often is customer data backed up, where are the backups kept, and when was a restore last tested?The backup schedule, the backup location, and the date and result of the last restore test.IT leadBackups in the same account or region as the live system, or no restore test on record. MVSP asks for all data to be backed up to a different location from where the application runs. Ask for the last restore test date and how long the restore took.Checked (date). Next: at renewal.
R2. What is your disaster recovery plan, what recovery time and data loss does it plan for, and when was it last tested?A summary of the plan with its recovery time and recovery point objectives, and the date and outcome of the last test.Operations leadObjectives with no test behind them. MVSP asks for disaster recovery plans to be tested at least annually or after significant changes, and the NCSC asks what plan a supplier has for maintaining minimum service levels to you if it suffers an incident such as a ransomware attack. Ask what happens to your service if the main hosting region is down for a day.Checked (date). Next: at renewal.
R3. What availability do you commit to, and where do you publish incidents and outages?The service level clause, the status page address, and the last twelve months of incident history.Operations leadA target with no remedy when it is missed, or no public history. Read the status history yourself and compare it with the target, then ask what remedy applied the last time the target was missed.Checked (date). Next: every quarter, from the status page.
R4. What happens to our data and access if your company is sold, stops trading or retires the product?The contract terms on change of control and end of service, the notice period, and the data return process.Contract ownerNo notice period for the end of service. Ask how much notice you would receive, and confirm that export (row H5) keeps working until the last day.Checked (date). Next: at renewal, or on news of a sale.
R5. At contract end, how will our data be returned and deleted, including backups, and how will you confirm it?The exit clause, the deletion timeframe for live data and backups, and a sample of the written confirmation of deletion.Contract ownerDeletion "under our standard practices" with no timeframe and no confirmation. The NCSC asks what the contract provides for secure deletion or return of your data at exit. Ask for written confirmation of deletion by a set date, backups included.Checked (date). Next: at contract end.

Question

Evidence to request

  1. G1. Who at your company is accountable for security, and who stands in for them when they are away?
    Evidence to request
    The names and roles of the accountable person and a deputy, taken from the security policy or an organisation chart.
    Answer owner (your side)
    Contract owner
    Red flag and next question
    "Our IT team" or "everyone" with no named person, or one founder who is also the only developer. Ask for the name and phone number your team would call if the product were breached at night.
    Review date
    Checked (date). Next: at renewal, or when either person leaves.
  2. G2. Which written security policies apply to the product, and when was each one last reviewed and approved?
    Evidence to request
    A list of policy titles with version numbers, owners and approval dates. The full documents under a confidentiality agreement, if the vendor's tier calls for them.
    Answer owner (your side)
    IT lead
    Red flag and next question
    Undated policies, a template with another company's name left in, or policies about an office network that say nothing about the product. Ask for the approval record of the current versions.
    Review date
    Checked (date). Next: at renewal.
  3. G3. Which independent certifications or audit reports cover this product, who issued them, and what scope and dates do they cover?
    Evidence to request
    The document itself: a SOC 2 report, an ISO/IEC 27001 certificate or a CSA STAR Level 2 listing, showing the issuer, the dates and the service it covers.
    Answer owner (your side)
    IT lead
    Red flag and next question
    A badge on the website instead of the document, or a certificate that covers a parent company, an office or a different service. The NCSC asks whether a certification's scope covers the parts of the service you will be using. Ask for the scope statement and the issue and expiry dates.
    Review date
    Checked (date). Next: (certificate expiry, or the end of the report period).
  4. G4. How can outsiders report a vulnerability in the product, and what happens to a report once it arrives?
    Evidence to request
    The published vulnerability disclosure policy, the security.txt file at /.well-known/security.txt if there is one, and the triage procedure with its response times.
    Answer owner (your side)
    Technical lead
    Red flag and next question
    No reporting route at all, or a contact address that bounces. Check the policy page and the file yourself. Ask when the last report arrived and how long it took to resolve.
    Review date
    Checked (date). Next: when the Expires date in the security.txt file passes, or at renewal.
  5. G5. How are staff and contractors who can reach customer data trained in security, and how is completion recorded?
    Evidence to request
    Training records showing role-specific content and completion dates for everyone with production or support access, contractors included.
    Answer owner (your side)
    IT lead
    Red flag and next question
    One generic video a year with no completion record, or contractors left out. MVSP asks for role-specific security training relevant to each person's business function. Ask for the completion rate among staff who can reach production.
    Review date
    Checked (date). Next: at renewal.
  6. G6. Which of the answers in this questionnaire will you commit to in the contract?
    Evidence to request
    The vendor's security schedule or data processing terms, marked against the rows of this questionnaire they cover.
    Answer owner (your side)
    Contract owner
    Red flag and next question
    Generous answers on the form that appear nowhere in the contract, or "our terms are fixed" for a full-tier service. Ask which rows they will put in writing, and treat the rest as unconfirmed.
    Review date
    Checked (date). Next: at each contract change or renewal.
  7. T1. When was the product last penetration tested, and what did the scope include?
    Evidence to request
    The executive summary and the scope as tested from the latest report: the application, the API, the roles, the environment, the dates and the exclusions.
    Answer owner (your side)
    Technical lead
    Red flag and next question
    A test of the marketing website, a scope that leaves out the API or the roles you will use, or a test date before the last major release. Ask for the scope page and the release history since the test.
    Review date
    Checked (date). Next: twelve months after the report date, or after the vendor's next major release.
  8. T2. Who carried out that test, and were they independent of the people who built and run the product?
    Evidence to request
    The testing provider's name, and a statement of whether it built, hosts or maintains the product.
    Answer owner (your side)
    Technical lead
    Red flag and next question
    The development agency reviewing its own work, or an internal team exercise described as independent. Ask how the tester relates to the builder. If your policy requires a CREST-accredited tester, ask for the provider's name and confirm the accreditation before you rely on the report.
    Review date
    Checked (date). Next: with T1.
  9. T3. Which critical and high-risk findings were reported, and what is the retest status of each one now?
    Evidence to request
    The findings table, with a severity for each finding and its retest status: fixed, partially fixed, still reproducible or not retested.
    Answer owner (your side)
    Technical lead
    Red flag and next question
    "All findings remediated" with no retest behind it, or a count of findings with no statuses. A finding marked not retested has not been shown to be fixed. Ask for the retest date and the build it ran on.
    Review date
    Checked (date). Next: (date the last open finding is due for retest).
  10. T4. Which methods or standards did the testing reference, and how are the findings mapped to them?
    Evidence to request
    The methodology section of the report, naming references such as the OWASP Web Security Testing Guide or ASVS 5.0, with each finding mapped to a category.
    Answer owner (your side)
    Technical lead
    Red flag and next question
    "Tested against the OWASP Top 10" offered as a pass mark. Naming a document describes how the testing was organised, not what it showed. Ask which checks were run for the roles you will use, and what the checks that found nothing covered.
    Review date
    Checked (date). Next: with T1.
  11. T5. How do you find and fix vulnerabilities in your code and third-party components between tests, and how fast?
    Evidence to request
    The patching timeframes by severity from the vendor's policy, and last quarter's dependency alerts with the date each one was closed.
    Answer owner (your side)
    Technical lead
    Red flag and next question
    "We run automated scans" with no timeframes or records. MVSP's time-to-fix control asks for patches to application vulnerabilities that materially impact security within 90 days of discovery. Ask for the vendor's own timeframes and its record against them.
    Review date
    Checked (date). Next: at renewal.
  12. T6. Can you provide a software bill of materials (SBOM), or another current list of the product's third-party components?
    Evidence to request
    An SBOM in a standard, machine-readable format, which CISA's Secure by Demand guide suggests asking for, or a dated list of components with their versions.
    Answer owner (your side)
    Technical lead
    Red flag and next question
    No inventory at all. OWASP Top 10:2025 says you are likely vulnerable to supply chain failures if you do not carefully track the versions of all the components you use, including nested dependencies. Ask how the vendor would learn that a component in the product had a published vulnerability.
    Review date
    Checked (date). Next: at renewal.
  13. T7. Will you allow security testing of the product commissioned by a customer, and on what terms?
    Evidence to request
    The contract clause or published testing terms, naming the environment, the notice required and anything excluded.
    Answer owner (your side)
    Technical lead
    Red flag and next question
    A flat refusal from a vendor that has no recent independent test to offer instead. MVSP's customer testing control asks vendors to let customers or their delegates test on request, on production or on a non-production environment that closely resembles it in functionality. Ask for written terms before anyone sends a single request.
    Review date
    Checked (date). Next: before any test is commissioned.
  14. H1. In which countries is customer data stored, processed and backed up, by which hosting providers, and from where do support and development staff reach it?
    Evidence to request
    A list of hosting providers and regions for production data and backups, and the countries that support and development staff work from.
    Answer owner (your side)
    Privacy lead
    Red flag and next question
    "In the cloud" with no region, or storage in Australia with backups or support handled elsewhere. The NCSC asks whether a supplier offshores any part of its service, including support, development or maintenance, and whether it will tell you if locations change. Ask for notice before any location changes.
    Review date
    Checked (date). Next: at renewal, or when the vendor announces a new region or hosting provider.
  15. H2. How is customer data encrypted in transit and at rest, including backups, and who manages the keys?
    Evidence to request
    The vendor's security overview or architecture note naming the encryption used for connections, databases, file storage and backups, and who holds the keys.
    Answer owner (your side)
    IT lead
    Red flag and next question
    "Industry-standard encryption" with no detail, or encryption at rest that stops short of backups and file storage. Ask which data stores the answer covers, one by one.
    Review date
    Checked (date). Next: at renewal.
  16. H3. Which types of data will the product hold for you, and do you keep a current diagram of how that data flows?
    Evidence to request
    A list of the data types the product processes, and a data flow diagram showing where each one enters and where it is stored.
    Answer owner (your side)
    Privacy lead
    Red flag and next question
    The vendor cannot say what it holds or where it ends up. MVSP lists both a list of sensitive data types and an up-to-date data flow diagram among its controls. Ask for the diagram covering the features you will switch on.
    Review date
    Checked (date). Next: when you start using a new feature or sending a new type of data.
  17. H4. How is one customer's data kept apart from another customer's inside the product, and was that separation tested?
    Evidence to request
    The architecture description of how customer accounts are separated, and the findings or negative results from test checks run between two customer accounts.
    Answer owner (your side)
    Technical lead
    Red flag and next question
    "Every customer has their own login" offered as separation. Ask whether the last test used accounts in two customer organisations, and what it found.
    Review date
    Checked (date). Next: with T1.
  18. H5. Can we export all of our data in a usable format, at any time and at contract end?
    Evidence to request
    The export documentation, a sample export, and the contract clause on returning data.
    Answer owner (your side)
    Operations lead
    Red flag and next question
    Export only through a support request, only for some records, or in a format nothing else can read. Run an export yourself during the trial and open the file, then ask how a full export of a large account is delivered and how long it takes.
    Review date
    Checked (date). Next: run a test export once a year.
  19. H6. Is production data ever copied into test, development or support environments?
    Evidence to request
    The policy on data in non-production environments, and a description of how test data is created or masked.
    Answer owner (your side)
    Privacy lead
    Red flag and next question
    Developers working on copies of live customer records. MVSP, in its customer testing control, says non-production environments should not contain production data. Ask how test data is made and who can reach any copies.
    Review date
    Checked (date). Next: at renewal.
  20. S1. Which third parties can access customer data, what does each one do, and where is it based?
    Evidence to request
    The current sub-processor list, naming each company, its role and its location.
    Answer owner (your side)
    Privacy lead
    Red flag and next question
    Categories such as "cloud provider" instead of company names, or no list at all. MVSP asks vendors to keep a list of third-party companies with access to customer data and make it available to clients on request. Ask for the named companies.
    Review date
    Checked (date). Next: at renewal, and on each change notice.
  21. S2. How will you tell us before adding or replacing a sub-processor, and can we object?
    Evidence to request
    The notification clause, the notice period and the channel used, such as an email list or a web page with a change history.
    Answer owner (your side)
    Contract owner
    Red flag and next question
    Changes made quietly on a web page with no notice. The NCSC asks whether you would wish to be informed if a supplier changes subcontractors. Ask for notice before the change takes effect, and a route to object or to leave.
    Review date
    Checked (date). Next: on each change notice.
  22. S3. How do you assess the security of your own suppliers, and when did you last review your main hosting provider?
    Evidence to request
    The vendor's supplier review procedure, and the date and outcome of its last review of its two most important sub-processors.
    Answer owner (your side)
    IT lead
    Red flag and next question
    "Our hosting provider handles security" with no review record. MVSP asks vendors to assess third parties with access to customer data every year, against the latest MVSP release. Ask for the last review dates.
    Review date
    Checked (date). Next: at renewal.
  23. S4. Which of your commitments in this contract are passed down to the sub-processors that hold our data?
    Evidence to request
    The flow-down clauses in the vendor's sub-processor agreements, or a summary of them.
    Answer owner (your side)
    Contract owner
    Red flag and next question
    Commitments on incident notice or data location that the vendor's own suppliers are not bound by. The NCSC asks whether controls in your contract with a supplier must be passed down to all subcontractors. Ask which commitments flow down, row by row.
    Review date
    Checked (date). Next: at renewal.
  24. S5. Which AI services or other outside processors receive our data through features of the product, and can that data be used to train models?
    Evidence to request
    A list of features that send data to an outside AI or processing service, the provider behind each one, the data it receives, and the setting or contract term on model training.
    Answer owner (your side)
    Privacy lead
    Red flag and next question
    AI features switched on by default with no list of providers, or a general line about not selling data in place of an answer on training. Ask whether each feature can be switched off for your account, and how.
    Review date
    Checked (date). Next: whenever the vendor announces a new AI feature.
  25. A1. Do you support single sign-on with our identity provider and multi-factor sign-in for every user, and on which plans?
    Evidence to request
    The product documentation for single sign-on and multi-factor sign-in, the plans that include each, and the settings as seen in a trial account.
    Answer owner (your side)
    IT lead
    Red flag and next question
    Single sign-on only on the top plan, multi-factor sign-in that is optional and off by default, or a shared default password at setup. CISA's Secure by Demand guide asks whether a manufacturer supports standards-based single sign-on at no additional cost and, where it manages authentication, enables multi-factor authentication by default at no cost. Ask whether administrators can enforce it for every user.
    Review date
    Checked (date). Next: at renewal, and whenever you change plan.
  26. A2. Which of your staff can access customer data, how is that access approved, and when was it last reviewed?
    Evidence to request
    The access policy, the number of staff with access to production data, and the date and outcome of the last access review.
    Answer owner (your side)
    IT lead
    Red flag and next question
    "Only authorised staff" with no number and no review record, or every support agent able to open every customer's account. MVSP asks vendors to limit sensitive data access to users with a legitimate need and to review access regularly. Ask whether staff access to your data is logged, and whether you can see that log.
    Review date
    Checked (date). Next: at renewal.
  27. A3. Is multi-factor authentication required for staff who reach production systems or customer data remotely?
    Evidence to request
    The identity provider's enforcement setting and the list of systems it covers: hosting console, database, code repository and support tools.
    Answer owner (your side)
    IT lead
    Red flag and next question
    Multi-factor sign-in on email only, with the hosting console or support tools left on passwords. MVSP asks that remote access to customer data or production systems require multi-factor authentication. Ask for the list of systems.
    Review date
    Checked (date). Next: at renewal.
  28. A4. How quickly is access removed when a staff member or contractor leaves?
    Evidence to request
    The leaver procedure with its timeframe, and a recent leaver record with the name removed, showing the date access ended in each system.
    Answer owner (your side)
    IT lead
    Red flag and next question
    No timeframe, or contractors handled outside the process. MVSP asks for redundant accounts and expired access grants to be deactivated in a timely manner. Ask for the most recent leaver record.
    Review date
    Checked (date). Next: at renewal.
  29. A5. What does the product log about sign-ins, administrator actions and data access, how long are those logs kept, and can we export them?
    Evidence to request
    The audit log documentation, the retention period, the plans that include it, and a sample log screen from a trial account.
    Answer owner (your side)
    IT lead
    Red flag and next question
    Audit logs only on the top plan, or kept for a few days. CISA's Secure by Demand guide says cloud and SaaS providers should retain security logs and make them available to customers for at least six months at no additional charge, and MVSP sets 30 days at no additional charge as its minimum. Ask whether the vendor's own staff access to your data appears in the log.
    Review date
    Checked (date). Next: at renewal.
  30. A6. How does the product control what each role inside our account can see and change?
    Evidence to request
    The role and permission table from the documentation, checked in a trial account with two different roles.
    Answer owner (your side)
    Technical lead
    Red flag and next question
    Only an administrator role and a user role, so anyone who needs one setting gets all of them. Ask whether permissions can be limited by team, by record type or by action.
    Review date
    Checked (date). Next: when you add a team or a new kind of user.
  31. I1. How quickly will you notify us of a security incident affecting our data, through which channel, and what will the notice contain?
    Evidence to request
    The contract clause with the notice period in hours or days, the channel and the named contacts, and the notice template if there is one.
    Answer owner (your side)
    Contract owner
    Red flag and next question
    "Without undue delay" with no number, or notice only through a public status page. MVSP's incident handling control asks vendors to notify relevant parties of a breach affecting sensitive information no later than 72 hours after discovery, with its nature, contacts, consequences and the measures taken. Ask for a number in the contract and a named contact on each side.
    Review date
    Checked (date). Next: at renewal, and after any incident.
  32. I2. Have you had a security incident or breach affecting customer data in the last three years, and what changed afterwards?
    Evidence to request
    A summary of each incident: the date, what happened, the data affected, which customers were told, and the changes made since.
    Answer owner (your side)
    Contract owner
    Red flag and next question
    "None" from a vendor whose own announcements say otherwise, or an incident followed by no change. Search the vendor's name with "breach" and "incident" before you accept a none. Ask what changed and how they checked that it worked.
    Review date
    Checked (date). Next: at renewal.
  33. I3. Do you have a written incident response plan, and when was it last exercised?
    Evidence to request
    The plan's contents page and its owner, and the date and lessons of the last exercise or real incident.
    Answer owner (your side)
    IT lead
    Red flag and next question
    A plan with no step for telling customers, or one that has never been exercised. Ask what the last exercise found and what was changed because of it.
    Review date
    Checked (date). Next: at renewal.
  34. I4. Who will our team work with during an incident, and how will you include us in the response?
    Evidence to request
    Named incident contacts with phone numbers, and the vendor's process for working with customers while an incident is under way.
    Answer owner (your side)
    IT lead
    Red flag and next question
    A general support inbox as the only contact. Ask for a phone number that is answered out of hours, and add it to your own incident contact sheet.
    Review date
    Checked (date). Next: every six months, by checking the number still answers.
  35. I5. After an incident, what will you share with us: logs, a timeline, a root cause report?
    Evidence to request
    The contract or policy term on post-incident reports, and on keeping logs long enough for an investigation.
    Answer owner (your side)
    IT lead
    Red flag and next question
    No commitment beyond a status page update, or logs deleted on a short cycle. Ask how long logs are kept (row A5) and whether affected customers receive a written root cause report.
    Review date
    Checked (date). Next: at renewal.
  36. R1. How often is customer data backed up, where are the backups kept, and when was a restore last tested?
    Evidence to request
    The backup schedule, the backup location, and the date and result of the last restore test.
    Answer owner (your side)
    IT lead
    Red flag and next question
    Backups in the same account or region as the live system, or no restore test on record. MVSP asks for all data to be backed up to a different location from where the application runs. Ask for the last restore test date and how long the restore took.
    Review date
    Checked (date). Next: at renewal.
  37. R2. What is your disaster recovery plan, what recovery time and data loss does it plan for, and when was it last tested?
    Evidence to request
    A summary of the plan with its recovery time and recovery point objectives, and the date and outcome of the last test.
    Answer owner (your side)
    Operations lead
    Red flag and next question
    Objectives with no test behind them. MVSP asks for disaster recovery plans to be tested at least annually or after significant changes, and the NCSC asks what plan a supplier has for maintaining minimum service levels to you if it suffers an incident such as a ransomware attack. Ask what happens to your service if the main hosting region is down for a day.
    Review date
    Checked (date). Next: at renewal.
  38. R3. What availability do you commit to, and where do you publish incidents and outages?
    Evidence to request
    The service level clause, the status page address, and the last twelve months of incident history.
    Answer owner (your side)
    Operations lead
    Red flag and next question
    A target with no remedy when it is missed, or no public history. Read the status history yourself and compare it with the target, then ask what remedy applied the last time the target was missed.
    Review date
    Checked (date). Next: every quarter, from the status page.
  39. R4. What happens to our data and access if your company is sold, stops trading or retires the product?
    Evidence to request
    The contract terms on change of control and end of service, the notice period, and the data return process.
    Answer owner (your side)
    Contract owner
    Red flag and next question
    No notice period for the end of service. Ask how much notice you would receive, and confirm that export (row H5) keeps working until the last day.
    Review date
    Checked (date). Next: at renewal, or on news of a sale.
  40. R5. At contract end, how will our data be returned and deleted, including backups, and how will you confirm it?
    Evidence to request
    The exit clause, the deletion timeframe for live data and backups, and a sample of the written confirmation of deletion.
    Answer owner (your side)
    Contract owner
    Red flag and next question
    Deletion "under our standard practices" with no timeframe and no confirmation. The NCSC asks what the contract provides for secure deletion or return of your data at exit. Ask for written confirmation of deletion by a set date, backups included.
    Review date
    Checked (date). Next: at contract end.

What a vendor security questionnaire is for

A vendor security questionnaire is the set of questions your business sends a supplier before it trusts that supplier with data, with access to your systems, or with part of how you serve customers.

This page is written for the buyer: the business choosing a SaaS platform, a booking or payments system, a web agency or a hosting provider, and deciding how much it is prepared to rely on them. If you are the supplier and a customer has sent you one, the guide to answering security questionnaires is the page for you.

The questionnaire is one step in a vendor security assessment, not the whole of it. The assessment decides how much the vendor matters to you, asks the questions that fit, checks the evidence behind each answer, writes any conditions into the contract and sets a date to look again.

NIST's Cybersecurity Framework 2.0 describes the same arc in its supply chain outcomes: suppliers are known and prioritised by criticality (GV.SC-04), planning and due diligence are performed to reduce risks before a formal supplier relationship begins (GV.SC-06), and the risks a supplier poses are monitored over the course of the relationship (GV.SC-07).

You do not have to write the questions from nothing. Public bodies and industry groups publish question sets you can borrow from, and the forty below draw on the same ground:

  • The UK National Cyber Security Centre (NCSC) publishes supplier assurance questions, which it describes as questions to ask your suppliers that will help you gain confidence in their cyber security. They run from security governance and incident recovery through offshoring, personnel and physical security to independent testing and contract terms.
  • The US Cybersecurity and Infrastructure Security Agency (CISA) publishes a Secure by Demand guide with questions that organisations buying software can use to better understand a software manufacturer's approach to cybersecurity, covering sign-in, logs, software bills of materials and vulnerability disclosure.
  • The Cloud Security Alliance (CSA) 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 CSA's Cloud Controls Matrix.
  • Minimum Viable Secure Product (MVSP) describes itself as a list of essential application security controls. Its site lists procurement among the places to use it and says it is designed to be brief enough to include in request for proposal documents.

A completed questionnaire is a claim, not proof

A vendor that completes your questionnaire has told you what it says about itself.

That has value, because a written answer is a record of what the vendor told you, which you can ask to have written into the contract (row G6). It does not show that the product is secure, or that the vendor complies with any standard or law.

Ticking yes beside "Do you encrypt data at rest?" costs nothing. Producing the setting, the architecture note and the date it was last checked takes a little effort, and that effort is what separates a vendor doing the work from one filling in forms.

Published schemes draw the same line. CSA runs its STAR programme at two levels: Level 1 is a self-assessment, in which the provider submits its CAIQ, and Level 2 is a third-party audit, such as an ISO/IEC 27001 certification or a SOC 2 attestation, combined with a CAIQ submission.

A CAIQ on its own is the vendor's own account, so check which level a listing sits at before reading anything into it.

CISA's guide points to a second gap. It notes that due diligence on software manufacturers often focuses on how they protect their own infrastructure and operations, such as whether they meet various compliance standards, and that relatively few of the compliance standards used in procurement focus on product security: what a manufacturer does so that the product it delivers holds up against attackers.

A vendor can hold a certificate for its office systems and still ship a product with weak sign-in, coarse roles and thin logs.

That is why many of the forty questions are about the product you will use, not the company.

Grade the evidence, not the answers

When the replies come back, sort the evidence behind each answer into four grades, strongest first. A row is only as strong as the best grade behind it.

  • Checked by you. Things you can confirm without trusting anyone: the sign-in options in a trial account, the audit log screen, a published vulnerability disclosure policy, a sub-processor page, a status page with its incident history, an export you ran and opened yourself.
  • Independent and specific. A report or certificate from an outside party that names the product you are buying and carries dates: a SOC 2 report, an ISO/IEC 27001 certificate whose scope covers the service, or the summary, scope and retest status of a recent penetration test. The AICPA describes SOC reports as assurance reports CPAs can provide, giving users the information they need to assess and address the risks of outsourced services.
  • The vendor's own records. Policies, plans, diagrams, access reviews and restore test results. They show intent and, when dated, practice, but nobody outside the vendor has checked them.
  • A bare answer. Yes, no, or a paragraph with nothing behind it. Treat it as an open item until something higher on this list supports it.

An honest gap can outweigh a confident yes

"Not yet; single sign-on reaches your plan in March, and the contract can say so" tells you the vendor knows where its gaps are and is willing to be held to a date. An unsupported yes tells you only that someone was willing to tick a box. Weigh the first more heavily, and write the date into the contract.

How to send the questionnaire and collect the answers

Work out the vendor's tier first, as the next section shows, and send only the rows that tier needs.

Each row in the template carries five things:

  • The question, with a reference such as G1 or T3, so answers and follow-ups can be matched without retyping them.
  • The evidence to request. Ask for it with the answer, not afterwards, and say that a document may be shared under a confidentiality agreement or shown on a call instead of sent.
  • The answer owner: the person on your side who reads that answer and decides whether it is good enough. The template uses five roles. The contract owner runs the relationship and signs. The IT lead is your in-house IT person or managed IT provider. The privacy lead handles personal information questions. The technical lead understands the integrations and the application. The operations lead knows what stops when the service does. In a small business one person may hold three of these roles; write the name in anyway, so nobody assumes someone else read the answer.
  • The red flag and next question: what a weak answer can look like, and the follow-up that tells you whether it is weak or just badly worded.
  • The review date: the day the answer was checked, and the event or date that should trigger the next check.

Collecting the replies

Give the vendor a reply date and a named person to direct questions to. Ask it to name who answered each section, because an answer written by a salesperson and one written by an engineer read differently, and you may need to go back to the engineer.

If the vendor sends its standard security pack instead, accept the pack as evidence, then map it against your rows. Whatever the pack does not cover stays open, and you can ask those rows on their own.

Keep the completed sheet with the signed contract. It records what the vendor told you when you decided to buy, and it is the starting point for every review after.

Scaling the forty questions to the vendor's risk

Sending all forty questions to the maker of a design tool that only ever sees your marketing images wastes time on both sides, and the answers that matter get lost among the ones that do not.

Before you send anything, answer three questions about the vendor:

  • What will it hold or see? Nothing sensitive; customer or staff contact details and ordinary business records; or sensitive personal information, payment flows, health or financial records, or credentials to your other systems.
  • What stops if it is gone for a week? Nothing much; a process you can run by hand for a while; or a service your customers use directly.
  • What can it do on your behalf? Nothing; read your data through an integration; or change data, send email as you, take payments or sign in to your systems.

Light tier: eight questions

For a vendor that holds nothing sensitive, could be replaced within a week and acts on nothing: G3, H1, H5, S1, A1, I1, R1 and R5. Together they cover any independent evidence, where your data sits, who else sees it, how people sign in, whether you will hear about an incident, backups, and whether you can leave with your data. Documentation links are acceptable evidence at this tier. Check A1 yourself in a trial account.

Standard tier: twenty-three questions

For a vendor that holds customer or staff contact details or ordinary business records, or that a team relies on every day: the eight light-tier rows plus G1, G2, G4, T1, T3, H2, H3, S2, A2, A3, A4, A5, I2, I3 and R2. Ask for evidence on every row, and accept the vendor's own dated records where no independent report exists.

Full tier: all forty, plus a call

For a vendor that holds sensitive information, handles payments, runs a service your customers use directly, or can act inside your systems: every row, independent evidence wherever a row asks for it, and a call with the vendor's technical lead to walk through T1 to T7 and H4. A row that cannot be answered becomes a contract condition with a date, or the reason to choose someone else.

The highest answer to any of the three questions sets the tier. One vendor can also move between tiers: a scheduling tool that starts with staff names becomes a different risk the day you connect it to your customer records.

Weighing the replies: accept, attach conditions or decline

Once the answers are in, mark each row with one of four results and keep the mark next to the evidence:

  • Evidenced:the answer is backed by something from the top two evidence grades and covers the product you are buying.
  • Stated:a reasonable answer with only the vendor's word, or its own undated documents, behind it.
  • Gap:the vendor says it does not do this yet. Note whether it gave a date.
  • Evasive:the reply does not answer the question asked, points to a marketing page, or answers a different question.

Patterns that repeat in weak replies

A few Evasive marks on a full-tier vendor say more than any single answer. Watch for these patterns in a reply: "we take security seriously" in place of a fact; "industry-standard" with no detail; a certificate for a parent company or a different product; a penetration test of the marketing site offered for questions about the application; "all findings remediated" with no retest; a refusal to name sub-processors; and an incident history of "none" from a vendor whose own blog announced an incident.

Turning the marks into a decision

Decide row by row, then overall: accept; accept with conditions written into the contract; or decline. CISA's guide places product security across the whole purchase, with questions before procurement, product security requirements in contract language during it, and continued assessment after it. CSF 2.0 outcome GV.SC-05 makes the same point about contracts: requirements to address supply chain risks are integrated into contracts and other agreements with suppliers.

The conditions worth writing down usually come straight from the red flag column: an incident notice period stated in hours, notice before a sub-processor changes, a date for single sign-on on your plan, a retest of an open finding, and the export format at exit. Give each condition a date and an owner on both sides, so it is not left to memory.

Reading a vendor's penetration test evidence

Rows T1 to T4 ask for test evidence, and this is where a vendor can say a great deal while showing very little.

A penetration test report describes one system, as it stood during one testing window, inside a scope the vendor agreed with its tester. It is not a certificate, and a vendor that presents it as one is overstating it. What to ask for, and what to check:

  • The right extract. The executive summary, the scope as tested and the findings table with retest statuses answer most buyer questions. You do not need the reproduction steps, and a vendor is right to hold those back or share them only under a confidentiality agreement.
  • The product and the roles. The scope should name the application you will use, its API, and the roles your staff will be given. A test of the vendor's marketing site says nothing about the application behind the login.
  • The date against the release history. A test that predates the vendor's last major release describes the product as it was then, not as it is now.
  • The tester. Ask whether the tester built, hosts or maintains the product. A builder's review of its own work is useful to the builder and is not independent evidence. If your own policy requires a CREST-accredited tester, ask for the provider's name. CREST awards accreditation to organisations after an application, evidence submission and review, and accredited companies are reassessed periodically.
  • The ratings and the statuses. Each finding should carry a severity, such as a CVSS v3.1 score on its 0.0 to 10.0 scale, and a retest status: fixed, partially fixed, still reproducible or not retested. A finding marked not retested has not been shown to be fixed, whatever the questionnaire answer says.
  • The references. A report may cite the OWASP Web Security Testing Guide or ASVS 5.0, or map findings to the OWASP Top 10:2025, which OWASP describes as an awareness document on the most critical security risks to web applications. Citing a document describes how the work was organised. It is not a pass mark.

How often a vendor should be testing

MVSP's external testing control asks vendors to contract a security vendor to perform comprehensive penetration tests of their products, services and dependent systems at least annually. That makes a twelve-month-old report a reasonable point to ask for the next one, and a much older report a reason to treat the testing rows as Stated rather than Evidenced. Our sample penetration test report, built from a lab application, shows the format.

Keeping the answers current after you sign

An answer is true on the day it was given.

CSF 2.0 outcome GV.SC-07 asks for supplier risks to be monitored over the course of the relationship, and the review date column is where that happens in practice. Each row suggests a trigger as well as a date, because an answer can go stale through an event as well as with time:

  • Renewal. Send the rows for the vendor's tier again, starting from last year's answers, and ask only what has changed.
  • A change in what you send the vendor. The NCSC asks how you will monitor for changes in a supplier's information risk profile over time, for example when the volume of data it processes significantly increases, when different types of data are introduced, or when new technology is introduced. The usual result is moving the vendor up a tier.
  • A change at the vendor: a new sub-processor, a new hosting region, a new AI feature, a sale of the business, or a security incident.
  • An expiry: a certificate or audit report period ending, a test report turning a year old, or the Expires date in the vendor's security.txt file passing. RFC 9116 recommends that date be less than a year into the future, so a stale file is itself a small signal.

Plan the exit before you need it

CSF 2.0 outcome GV.SC-10 expects supply chain risk management plans to cover what happens after a partnership or service agreement ends, and the NCSC asks what the contract provides for secure deletion or return of your data at exit, including transfer to another supplier.

Rows H5, R4 and R5 cover it. Run a test export once a year, so the first time you try is not the week you need it.

Put key vendors in your own incident plan

CSF 2.0 outcome GV.SC-08 includes relevant suppliers in incident planning, response and recovery. In practice, the contact from row I4 belongs on your own incident contact sheet, and you should know which vendor you would call first for each kind of outage.

Where a security test fits in a vendor review

Clever Ops tests what customers reach through a browser or an API: websites, WordPress and WooCommerce sites, custom web applications and their APIs, apps made with AI builders, and AI chatbots and agents.

In a vendor review, that matters in two situations.

  • A website or web application is being built for you. Once it is yours, you can have it tested before launch instead of relying on the builder's questionnaire answers. No request is sent until the owner of each target has signed an authorisation and the hosting provider's testing policy has been checked. If Clever Ops built or hosts it, the work is a security review rather than an independent test, and the report says who built and who hosts it.
  • A software vendor has no recent test of the application you will use. The vendor can commission one, and the vendor has to authorise it, because the vendor owns the system. CISA's vulnerability disclosure policy template, written to help US federal agencies create disclosure policies, says in its scope section to confirm whether a SaaS vendor has explicitly authorised testing, for example in the contract or its public policy, to work with the vendor to obtain authorisation if not, and to leave the system out of scope if authorisation cannot be obtained. RFC 9116 adds that researchers should not assume a security.txt file, present or absent, grants or denies permission for security testing.

What a test cannot stand in for

Several questionnaire rows ask for evidence no security test produces. Clever Ops is not ISO 27001 certified and does not issue SOC 2 reports. Clever Ops is not a Qualified Security Assessor or an Approved Scanning Vendor. For a vendor that handles card payments for you, ask for its own PCI DSS validation documents: Qualified Security Assessor companies are qualified by the PCI Security Standards Council to validate an entity's adherence to PCI DSS, and an Approved Scanning Vendor conducts external vulnerability scanning to validate adherence with the external scanning requirements of PCI DSS Requirement 11.3.2.

Where your own policy requires CREST, Clever Ops is not CREST accredited, so we refer that work to a provider that is.

Frequently asked questions

A set of questions a business sends a software, cloud or service supplier before trusting it with data, system access or part of its operations. A useful one asks for evidence with each answer rather than a bare yes or no, and is sized to how much the vendor matters. Depending on who sends it, you may see it called a supplier security questionnaire or a cyber security questionnaire.

The questionnaire is the list of questions. The assessment is the decision around it: placing the vendor in a risk tier, sending the questions that fit, checking the evidence, writing conditions into the contract and reviewing the answers on a schedule. A questionnaire filed without those steps records what a vendor said and little more.

No. It records what the vendor says about itself. What supports an answer is evidence: an independent audit report, a certificate whose scope covers the service you use, a recent penetration test summary with retest statuses, or a setting you checked yourself in a trial. Even that evidence speaks only for what it covered, on the dates it covered.

Often fewer than ten. If the supplier holds nothing sensitive, could be replaced quickly and cannot act inside your systems, the eight light-tier rows cover where the data sits, who else sees it, sign-in, incident notice, backups and leaving with your data. Sending all forty to a low-risk supplier costs both sides time and buries the answers that matter.

Ask for it if the vendor is a cloud provider that has one, because it saves both sides time. Check the vendor's STAR listing: Level 1 is a self-assessment and Level 2 adds a third-party audit. CAIQ answers are Yes or No, so pair them with the rows here that ask for evidence, especially T1 to T4 on testing of the product you will use.

Accept the pack as evidence and map it to your rows; whatever it does not cover stays open. A vendor that will not answer even the light-tier rows for a service holding your data has told you something useful. With a large platform that will not negotiate, lean on what you can check yourself and on its independent reports, limit the data you send it, and keep a current export.

Only with the vendor's written permission, because the vendor owns the system. Ask whether its contract or published testing terms allow customer testing, and on which environment. MVSP's customer testing control suggests vendors let customers or their delegates test on request. If permission is not given, ask for the vendor's own recent test evidence instead.

At each renewal at the latest, and sooner when something changes: a new sub-processor or hosting region, a new type of data you send, an incident at the vendor, an expired certificate, or a test report more than a year old. The review date column records both the date and the trigger for each row.

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.