Skip to content
Cyber Security Firms

How to Answer a Security Questionnaire, Step by Step

A security questionnaire is not a form. It is a set of representations your buyer can check, your lawyers will inherit, and your next customer will compare against the last set you sent.

Intermediate21 min readUpdated
How to Answer a Security Questionnaire, Step by Step

It arrives from a procurement address nobody recognizes, with a spreadsheet attached and a due date five business days out. Two hundred rows, a Yes/No column, a Comments column, an Evidence column. It was forwarded to you because you are the closest thing the company has to a security person, and the note on top says the deal cannot move until this comes back. Nothing in the document tells you what a good answer looks like. Every row you get wrong becomes a sentence your employer is held to for the life of the contract.

Why Did You Receive a Security Questionnaire?

Because a customer is about to hand you data, and somebody on their side has to write down why that was a reasonable thing to do. The questionnaire is the artifact that vendor risk teams produce as evidence of their own diligence, which is why it arrives late in the sales cycle and why it cannot usually be talked away.

The volume of these requests tracks a real trend rather than a fashion in procurement. Verizon's 2026 Data Breach Investigations Report found third-party involvement in breaches up 60 percent, with breaches involving a third party now accounting for 48 percent of all breaches. Almost half of incidents now arrive through somebody's supplier, and the buyers sending you a spreadsheet are the ones who read that number and want to know whether you are the supplier in question. The mechanics of how that happens are the subject of a supply chain attack, and every question on the form maps back to one step in it.

For some buyers the request is not discretionary at all.

How to Answer a Security Questionnaire

Nine steps, in order. The first three take an hour and save most of the pain in the other six, which is exactly the hour most first-timers skip because the due date is already close.

  1. Confirm what is in scope before you answer anything. Which product, which environment, which data classes, which of your legal entities. A questionnaire answered for your whole company when the buyer is only using one product is longer, weaker and harder to defend at renewal.
  2. Find the human who sent it. Get a name, a mailbox and a deadline you have agreed to rather than inherited. Every later step depends on being able to ask a question and get an answer inside a day.
  3. Read the whole thing once without answering. You are looking for duplicates, for questions that contradict each other, for the two or three that will need engineering, and for anything that asks you to attest to something legal rather than technical.
  4. Sort every question into one of four types. Yes/no control questions, describe-your-process questions, evidence requests, and legal attestations. Each takes a different answer and a different approver, and mixing them is what produces essays in checkbox fields.
  5. Assign each section an owner and a date. Identity and access to whoever runs the identity provider, application security to engineering, personnel questions to HR, legal attestations to counsel. Give each one the deadline minus two days.
  6. Answer everything you can already evidence. Work the easy rows first so the remaining list is short and honest, and so you can see by day two how much genuine trouble you are in.
  7. Write the honest no's. For each gap: the plain no, the compensating control that reduces the risk today, and a remediation date that the person who has to deliver it has actually agreed to.
  8. Assemble and redact the evidence pack. Policies, the attestation report, the summary letter from your last test, the training completion export. Redact before you attach, not after somebody asks.
  9. Have one person read the whole document back before you send it. Consistency is what gets reviewed. Three people answering separately will contradict each other on retention, on subprocessors and on encryption, and the reviewer will notice.

Every one of those steps sits inside an exchange with the buyer that is longer than a single send and return, which is the part nobody warns you about.

The questionnaire round trip
BuyerYouQuestionnaire and due dateScope and clarification questionsScope confirmed, some questions withdrawnCompleted answers and evidence packFollow-ups on the partial answersClarifications and remediation dates
The second and fifth messages are the ones first-timers do not budget for, and they are where the calendar time goes.

What Does a Security Questionnaire Ask About?

The wording changes and the ground does not. Cloud Security Alliance organizes the same territory into 197 control objectives structured in 17 domains in its Cloud Controls Matrix, and most buyer questionnaires are a shorter walk over the same map.

DomainWhat the questions look likeWhat the buyer is testing
Governance and policyDo you have written policies, reviewed annually, approved by whomWhether security is somebody's job or everybody's afterthought
Identity and accessMFA coverage, least privilege, joiner and leaver process, access reviewsThe control set that fails most often in real incidents
Data protectionEncryption in transit and at rest, key management, retention, deletionWhether their data can be isolated and destroyed on request
Application securityCode review, dependency scanning, secure development, testing cadenceWhether the product itself is the attack surface
Infrastructure and operationsPatching clocks, logging, monitoring, configuration managementWhether anyone would notice an intrusion
Incident responsePlan, roles, tabletop exercises, notification timelines in the contractHow fast they would hear about it
Business continuityBackups, restore testing, recovery objectives, dependency on single regionsWhether an outage at your end stops their operation
PersonnelBackground checks, onboarding, security training completionThe insider and the untrained employee
Vendor and subprocessor managementWho you use, what they hold, how you assess themTheir fourth-party exposure, which is your third party

A question that does not fit any of these rows is usually specific to the buyer's own regulator, and it is the sort of thing worth asking about rather than guessing at.

Which Standard Security Questionnaires Will You See?

Which one lands on you says a lot about the buyer, how much room there is to negotiate, and whether an answer you write now will be reusable next quarter.

Question setWho publishes itWhat it signals
SIG and SIG-LiteShared Assessments, licensed to membersA buyer with a mature vendor risk function. SIG-Lite is the short form for lower-risk vendors
CAIQCloud Security AllianceA cloud buyer who will accept a published self-assessment instead of a bespoke form
Vendor SCRM TemplateCISA, free to downloadA public sector or critical infrastructure buyer, or a small team that wanted a credible starting point
NIST SP 800-171 basedDerived from the NIST publicationFederal or defense supply chain work, where the answers become a score rather than a comment
HECVATHigher education communityA university, usually with a shorter version available if you ask
The buyer's own spreadsheetWritten in-house, often years agoDuplicated questions, obsolete technology references, and the most room to negotiate scope

The free one is worth knowing about before a questionnaire arrives. CISA publishes a Vendor Supply Chain Risk Management Template developed by the ICT Supply Chain Risk Management Task Force, described as a set of questions about a supplier's implementation and application of industry standards and best practices. Answering it once, properly, gives you a draft answer for most of what any buyer will later ask, and it costs nothing.

A clipboard lying flat on a plain surface with a single sheet of paper clipped to it

How to Scope a Security Questionnaire Before You Answer It

Scope is the only lever that shortens the document, and it is available exactly once, before you start typing.

Ask the sender four things in one email. Which product and environment is the assessment covering. Which categories of their data will actually reach you. Whether questions outside that scope can be marked not applicable with a short reason. And whether they will accept an existing attestation report in place of the sections it already covers.

Mark a question not applicable only when you can finish the sentence "this does not apply because", and write that reason in the comment field. "N/A" alone reads as evasion and comes straight back as a follow-up. "Not applicable: the product is delivered as a hosted service and no customer data is processed on employee endpoints" reads as a company that understands its own architecture.

Resist the temptation to answer for the whole company. If the buyer is licensing one product, scope the answers to that product, its environment and the team that runs it. Broad scope makes every future answer harder to keep true, and a claim that quietly stops being true is worse than a no.

How to Write a Security Questionnaire Answer

An answer is not a sentence. It is four things stacked, and the reason reviewers send follow-ups is almost always that one of the four is missing.

What one good answer is made of
The claimYes, no or partial, in the wording the questionnaire uses
The controlThe named mechanism that makes the claim true
The scopeWhich systems, environments and data it covers
The evidence
The claim is what the reviewer reads first. The evidence is what a follow-up question is really asking for.

The four question types each want a different shape.

Yes/no control questions want the box, then one or two sentences in the comment. "Yes. Multi-factor authentication is enforced through our identity provider for all employee accounts and all production access, with no exemption group. Evidence: policy section 4.2 and the enforcement screenshot in the pack."

Describe-your-process questions want a short paragraph with a cadence and an owner in it. A process without a frequency and a named role is a paragraph about intentions. "Access reviews are performed quarterly by the system owner and countersigned by the head of engineering. Removals are actioned within five business days and recorded in the ticket."

Evidence requests want the document, not a description of the document. If you cannot attach it, say what it is, why it is restricted, and what you can offer instead.

Legal attestations want counsel, not you. Anything with the words indemnify, warrant, notify within, or comply with all applicable goes to whoever signs contracts, every time.

Write in the buyer's vocabulary rather than your own product's. If they say "data at rest" and you call it "the storage layer", translate. Encryption at rest is a phrase that maps to a row in their scoring tool, and an answer their tool cannot recognize gets flagged whatever it actually says.

How to Answer a Question When You Do Not Have the Control

This is the moment people freeze, and the wrong instinct is to soften a no into a maybe. A maybe fails twice: it does not satisfy the reviewer, and it costs you the credibility that makes your yes answers worth anything.

The pattern has three parts. The plain no. The compensating control that reduces the same risk today. A remediation date, if you are going to give one, that the person who has to deliver it has agreed to in writing.

"No. We do not currently run a formal bug bounty. Vulnerability reports reach a monitored security mailbox published in our security.txt, and external testing is performed annually by an independent firm. A bounty program is scheduled for Q2 2027, owned by the head of engineering."

Never invent the date. A remediation commitment made in a questionnaire tends to end up quoted in the contract, and a missed one is a breach of a representation rather than a slipped roadmap item.

Myth vs reality
What people believe
A single no will lose the deal, so soften it.
What actually happens
Reviewers expect gaps. They are scoring a pattern, not hunting for a perfect score, and most vendor risk teams have authority to accept a risk that is disclosed and dated. What loses deals is inconsistency: a yes on encryption in one section and a partial on key management in another, or a claim that contradicts the attestation report attached in the same email. A clean no with a compensating control is a smaller problem than a yes the reviewer decides not to believe.

How to Answer Questions About Your Cloud Provider

A large share of the infrastructure questions are about controls you do not operate. Physical security of the data center, hypervisor patching, media destruction: those belong to your hosting provider, and answering them as if they were yours is both wrong and impossible to evidence.

Attribute the control and then answer the part that is genuinely yours, which is oversight. "Physical and environmental controls are the responsibility of our infrastructure provider and are covered in its current attestation report, which we review annually as part of our subprocessor assessment. Our responsibility begins at the operating system layer, where we run configuration baselines and patching on a defined clock."

That second sentence is the one being marked. The buyer already knows who runs the data center. What they are testing is whether you know where the boundary is and whether anybody at your company looks at the provider's report.

If you are the cloud provider in this exchange, the reusable answer is a published self-assessment. Cloud Security Alliance's STAR Level 1 Security Questionnaire, CAIQ v4, is described as an industry-accepted way to document what security controls exist in IaaS, PaaS and SaaS services, as a set of yes/no questions mapped to the Cloud Controls Matrix. Publishing one to the STAR Registry means a share of buyers stop sending you a spreadsheet at all.

What Evidence Should You Send With a Security Questionnaire?

Attach the artifact, not the assertion. A policy PDF, an attestation report, a training completion export, a summary letter for the last independent test. Evidence is what converts a claim into something the reviewer can put in a file and defend later.

Redact before you attach. A summary letter from a penetration test is normal to share; the full report, with reproduction steps for findings that are still open, is a map of how to attack you and it will sit in a shared drive at a company you have no visibility into. Same rule for network diagrams, raw vulnerability scan exports, and anything naming individual employees. What a buyer needs is scope, date, tester, methodology, severity counts and remediation status.

Check what is covering the exchange before you send anything. Most buyers will sign a mutual NDA on request, and most reviewers expect to be asked.

How to Build a Security Questionnaire Answer Library

The second questionnaire should take a fraction of the first, and it will only do that if you kept the first one properly. A library is not the completed spreadsheets in a folder. It is a table of questions and approved answers that somebody owns.

The failure mode is a library that nobody re-reads. A control gets changed, an answer stays, and eight months later you send a buyer a statement about a system that no longer works that way. Tie the review to the same calendar as your access reviews so it happens whether or not a deal is waiting.

Can a SOC 2 Report Replace a Security Questionnaire?

Sometimes, partly, and it is always worth asking. A reviewer whose scoring tool has a row for each trust services criterion can often close most of the form from the report alone, and many will if you ask before you start typing.

What a report displaces is the section on controls it examined during the period it covers. A SOC 2 report is an accountant's opinion on controls over a stated window, which is stronger evidence than anything you write about yourself, and reviewers know it. What it does not displace is anything outside the audited scope, anything the auditor carved out to a subservice organization, and the complementary user entity controls, which are the things the report assumes the customer will do. Those come back as questions no matter how good the report is.

Ask this way: "We hold a current SOC 2 Type 2 covering security and availability for the product in scope. Would you like the report under NDA, and can we mark the sections it covers as evidenced by it?" A no costs you nothing. A yes can remove half the document.

What Happens If You Answer a Security Questionnaire Incorrectly?

Two things, in sequence. The answer becomes a representation in the contract, because vendor risk teams routinely attach the completed questionnaire as a schedule or reference it in the security addendum. Then somebody checks.

The public example of that mechanism is the federal one, and it is worth knowing even if you never sell to a government buyer, because the shape is the same everywhere. Under DFARS 252.204-7020, a defense contractor performs a Basic Assessment against NIST SP 800-171 and posts a summary score, which the clause itself describes as carrying a confidence level of low because it is self-generated. The government may then conduct a High Assessment, which includes verification, examination and demonstration of the contractor's system security plan to validate that the requirements were implemented as described. A self-assessment is an opening position, and the party receiving it has a documented route to checking whether it was true.

Commercial buyers do the same thing with softer instruments: an audit right in the contract, a request for the underlying evidence, a comparison against your attestation report, or the questions their own auditor asks them a year later. The exposure is not usually a regulator. It is a customer who relied on a sentence you wrote, and a breach that reaches data the sentence said was protected.

How Long Does a Security Questionnaire Take?

A short form from an organized buyer, answered by a company with a current library and a recent attestation report, is a morning. A two hundred question spreadsheet answered from scratch by a company that has never done one is two to four weeks of elapsed time, most of it spent waiting: for scope confirmation, for engineering to answer three questions, for counsel to look at the attestations, for the buyer to respond to the follow-up round.

The elapsed time is dominated by handoffs rather than typing, which is why the fix is the intake process and the named owners rather than faster writing. Companies that answer these well are not writing better prose. They are answering fewer questions, from a library, with evidence already redacted and waiting.

Key takeaways

  • A security questionnaire is the buyer's diligence record. For financial institutions under the FTC Safeguards Rule, periodically assessing service providers is a legal obligation, not a preference.
  • Scope it before you answer. Which product, which environment, which data. Scope is the only lever that shortens the document and it is available once.
  • Sort every question into four types: yes/no controls, describe-your-process, evidence requests and legal attestations. Each needs a different answer and a different approver.
  • A complete answer has four parts: the claim, the control, the scope and the evidence. Missing any one of them is what generates follow-up rounds.
  • Answer no plainly, with a compensating control and a date somebody has agreed to. Reviewers price disclosed gaps and punish inconsistency.
  • Attribute cloud provider controls to the provider and answer for your oversight of them, which is the part being scored.
  • Redact evidence before you send it. A summary letter is normal to share; a full test report with open findings is not.
  • Build the answer library after the first questionnaire, with an owner, an approval date and a review cadence, or the second one costs as much as the first.

Common questions

What is a security questionnaire?

A structured set of questions a prospective or existing customer sends to a vendor to assess how that vendor protects data. It usually covers governance, access control, data protection, application security, operations, incident response, continuity, personnel and subprocessors, and it forms part of the buyer's vendor risk record.

How long does it take to answer a security questionnaire?

A short form answered from an existing library is a morning. A two hundred question spreadsheet answered from scratch runs two to four weeks of elapsed time, most of it waiting on scope confirmation, engineering input, legal review of the attestations, and the buyer's follow-up round.

Can I mark questions as not applicable?

Yes, when you can finish the sentence "this does not apply because" and write that reason in the comment field. A bare N/A reads as evasion and returns as a follow-up. Agree the principle with the sender during scoping rather than deciding unilaterally halfway through.

What happens if I answer a security questionnaire incorrectly?

The completed questionnaire is commonly attached to or referenced by the contract, so a wrong answer becomes a representation you have made. Buyers verify through audit rights, evidence requests and comparison against your attestation report. If you find an error, correct it in writing promptly.

Do I have to answer every question?

Not necessarily. Questions outside the agreed scope, questions about controls owned by your hosting provider, and questions already covered by an attestation report can often be handled by reference rather than by a fresh answer. Get that agreed with the sender before you start.

What is the SIG questionnaire?

A standardized question set published by the Shared Assessments program and licensed to its members, widely used by financial services and other buyers with mature vendor risk functions. SIG-Lite is the shorter version generally used for lower-risk vendors.

What is CAIQ?

The Consensus Assessments Initiative Questionnaire from Cloud Security Alliance, a set of yes/no questions mapped to the Cloud Controls Matrix that a cloud consumer or auditor can ask of a provider. Providers can publish a completed CAIQ as a STAR Level 1 self-assessment so buyers read it instead of sending a form.

Can a SOC 2 report replace a security questionnaire?

Often it replaces part of one. A report is an independent opinion covering a period, which is stronger than a self-answered form, so many reviewers will close the sections it covers. What survives is anything outside the audited scope, anything carved out to a subservice organization, and the complementary user entity controls.

Should I attach our penetration test report?

Send the summary letter, not the full report. Scope, date, tester, methodology, severity counts and remediation status are what a reviewer needs. A complete report with reproduction steps for open findings is a set of instructions for attacking you, held by a company whose own security you cannot see.

Who should answer a security questionnaire?

One owner coordinating named contributors: identity and access to whoever runs the identity provider, application security to engineering, personnel questions to HR, legal attestations to counsel. One person then reads the whole document back before it goes, because consistency across sections is what gets reviewed.

How do I reduce the number of security questionnaires we receive?

Publish a trust page listing your certifications, subprocessors and document request process, keep a current attestation report available under NDA, and publish a completed standard questionnaire where your market uses one. Buyers with a documented alternative frequently take it.

On this page
Reviewed by

Daniel Reyes

  • CISSP
  • 12 years in security operations
  • Austin, TX

Daniel Reyes is a CISSP who spent twelve years in security operations, most recently leading a detection and response team for a mid-sized healthcare group in Texas. He reviews every resource and breach report on Cyber Security Firms for technical accuracy before it publishes.

Most of the people he has trained arrived having been told too much: a dozen acronyms, six vendors, and no clear idea which risk was theirs. His approach is to explain what an attack actually does before naming the tool that stops it, on the basis that most breaches start with something a reader could have recognised.

Read the full bio and how we research →