Buyer Guides · 6 MIN
A security questionnaire for tech vendors you can reuse
A practical security questionnaire for evaluating software vendors and AI providers, with what good answers look like and the standards each question maps to.
A useful vendor security questionnaire is short, specific, and mapped to known standards. Thirty questions across governance, access, data, development, AI use, incidents, and continuity will tell you most of what you need. This one is free to copy, with what a good answer looks like and the reference behind each group.
- Fewer, sharper questions beat a 400-line spreadsheet, and they get honest answers.
- Group questions by risk area and weight them by what the vendor will actually touch.
- Ask for evidence, such as a policy, a screenshot, or a report, not only a yes.
- Established references include AICPA SOC 2 reports, ISO/IEC 27001, NIST SSDF, and the OWASP lists.
- Add AI-specific questions if the vendor uses models, because standard questionnaires often skip them.
- Nactore is an AI-native technology and growth company and answers a questionnaire like this one for its own clients.
How should you use this questionnaire?
Send it as a document, ask for answers in writing, and request evidence for the items you weight highest. Adjust length to risk. A vendor who will only build a marketing site needs a light version. One who will access production data needs every section.
Score each answer 0 to 2, as in questions to ask a software agency before signing. Treat blanks and evasions as zeros, and follow up on any zero in a high-weight area.
What are the governance and access questions?
Standards behind this group include ISO/IEC 27001, which describes an information security management system, and the AICPA's SOC suite of services, under which CPAs issue assurance reports such as SOC 2.
- Do you have a written information security policy, and who owns it?
- Do you hold a SOC 2 report or ISO/IEC 27001 certificate? If so, share it and its scope.
- Is two-factor authentication required on email, code, cloud, and admin accounts?
- How do you grant, review, and remove access to client systems and data?
- Are staff under confidentiality obligations, and how are they trained on security?
- Which team members would access our data, and from where?
A good answer is specific. "Access is reviewed quarterly and removed within a day of departure" beats "we follow best practices".
What are the data protection questions?
- What categories of our data would you store or process?
- Where is it stored, and in which countries?
- Who are your sub-processors, including cloud, email, and AI providers?
- How is data encrypted in transit and at rest?
- How long do you retain our data, and how is it deleted at the end?
- Will you sign a data processing agreement? Under the EU GDPR, Article 28 sets the required contents for processors.
- How do you handle requests from users to access or delete personal data?
Which law applies depends on where your users are, so have counsel review the final terms.
What are the secure development questions?
The NIST Secure Software Development Framework (SP 800-218) and the OWASP Application Security Verification Standard are good references here.
- Is all code reviewed by a second person before release?
- Do you run automated tests and dependency vulnerability scans in your pipeline?
- How do you store and rotate secrets and API keys?
- Do you separate development, staging, and production environments?
- Do you test against the OWASP Top 10 weakness classes?
- How quickly do you patch critical vulnerabilities?
- Will our work run in our own accounts and repositories?
What are the AI-specific questions?
Standard questionnaires lag behind AI practice, so add these. The NIST AI Risk Management Framework and the OWASP Top 10 for LLM Applications are the references.
- Which AI tools and model providers will touch our code or data?
- Is our data used to train any model, and can you confirm the provider terms in writing?
- How is AI-generated code or content reviewed by a person before delivery?
- What can an AI agent do without human approval on our systems, and what is blocked?
- Are prompts, outputs, and agent actions logged, and for how long?
For more on controls, read LLM guardrails in production and MCP security risks.
What are the incident and continuity questions?
- What is your process when a security incident is suspected?
- How quickly will you notify us, and through which channel?
- Do you have backups, and when was a restore last tested?
- What happens to the project if a key person is unavailable?
- Do you carry cyber and professional liability insurance, and at what level?
Ask to see one thing, not thirty. Pick the item that matters most for your risk, such as the access removal checklist, and request the real document. Reality shows up fast.
How do you read the answers?
| Answer pattern | Likely meaning | Your move |
|---|---|---|
| Specific, with evidence | A real process exists | Score high, verify one item |
| General reassurance | Process may be informal | Ask for the document |
| "Not applicable" without reason | Misunderstanding or avoidance | Ask them to explain |
| Honest "no", with a plan | Gap, but candid | Decide if the gap fits your risk |
| Silence or delay | Weak practice or low priority | Treat as a warning |
Weigh candor positively. A small vendor that says "we do not hold SOC 2, here is what we do instead" is more trustworthy than one that stretches the truth. Small-vendor preparation is covered in how enterprises evaluate small tech vendors.
Frequently asked questions
Is a security questionnaire enough on its own?
For low-risk work, often yes, combined with contract terms. For high-risk access, add evidence review, a call with the vendor's technical lead, and possibly an independent assessment.
Which standard questionnaire should I use?
Many buyers use the Cloud Security Alliance CAIQ or the Shared Assessments SIG. The list above is a lighter alternative you can adapt, and it can be mapped to either.
How often should I repeat it?
At onboarding, at renewal, and after a significant change such as a new subprocessor or a security incident.
Who should review the answers?
Someone with security or engineering knowledge, plus legal for data and liability terms. Smaller companies can ask a trusted technical adviser to review the key sections.
The takeaway
A short, evidence-backed questionnaire reveals more than a long unread one. Copy the thirty questions, weigh them by risk, and verify the items that matter. Nactore is an AI-native technology and growth company that builds, ships, ranks, and grows as one partner, and we answer questionnaires like this one before work begins.
Want this built for your team? Book a free 30-minute call.
Want to apply this to your business?
Book a free 30-minute call. We will tell you what we would do first.