← All posts
SOC 2 audit exceptionscan you fail a SOC 2 auditSOC 2 qualified opinion

SOC 2 Audit Exceptions: Can You Fail a SOC 2 Audit?

A SOC 2 report has no pass or fail grade. Learn what exceptions are, when they change the opinion, how to respond, and how to keep them out of the next period.

An exception in a SOC 2 report is a recorded test result. Track each one from the auditor's test to verified remediation.
An exception in a SOC 2 report is a recorded test result. Track each one from the auditor's test to verified remediation.

You cannot fail a SOC 2 audit in the pass or fail sense, because the CPA firm issues an opinion instead of a grade. What you can get is a report with exceptions, which are test results where a control did not work as described, and in worse cases a modified opinion: qualified, adverse, or a disclaimer. A few exceptions with a clean opinion are common and survivable. A modified opinion is the result that hurts sales. This guide explains how exceptions turn into opinions, what to write when you get one, and how to keep the same exception out of your next period.

TL;DR

  • A SOC 2 examination ends in an opinion, not a grade. Unmodified is the clean result.
  • An exception is one test result. The CPA firm decides whether the exceptions, taken together, change the opinion.
  • Customers read the listed exceptions too, so a clean opinion with exceptions still draws questions.
  • Answer each exception with dates, root cause, and a fix you can prove. Do not argue the facts in the report.
  • Controls that run late, run without proof, or miss part of the population produce exceptions you could have caught. Test your own controls first.

What does it mean to fail a SOC 2 audit?

A SOC 2 examination is an attestation engagement under the AICPA attestation standards. The AICPA SOC 2 guide covers how the CPA firm plans the work, tests controls, and forms its opinion on the system description and the controls. The firm can issue one of four results:

Opinion What it means What a customer sees
Unmodified The description is fairly presented and the controls were suitably designed, and for Type 2, operated effectively. The clean result. Exceptions may still be listed in the tests section.
Qualified A material problem affects part of the subject matter, but not all of it. A paragraph naming the problem area. Expect questions from security teams.
Adverse The problems are material and pervasive. The report does not support reliance on your controls.
Disclaimer of opinion The CPA firm could not get enough evidence to form an opinion. No conclusion at all. Customers cannot rely on it.

“Material” and “pervasive” are the CPA firm’s professional judgments, not percentages you can compute in advance. That is why the honest answer to “how many exceptions can we have?” is “ask your CPA firm how it evaluates them,” not a number from a blog post.

Missing evidence can change the opinion as much as a broken control. If logs expired, a ticket system was migrated without an export, or a population cannot be reconciled, the firm may have nothing to test. That can lead to a qualified opinion for the affected area or, when the gap is widespread, a disclaimer. Losing the proof can hurt you as much as skipping the control.

What is a SOC 2 exception?

An exception, or deviation, is a single test result where a control did not operate as described. In a Type 2 report, the section describing the CPA firm’s tests of controls lists each test and its result. A typical entry reads like this:

Control Test performed Result
Access is removed within one business day of a worker’s departure. Sampled 25 departures and compared end date to removal time. Exceptions noted. 2 of 25 departures had access removed 4 and 9 days after the end date.

That row does not mean the report failed. It means a reader can see that the access removal control slipped twice in the period. Whether those two misses affect the opinion depends on the CPA firm’s evaluation of the control, the risk, and any other controls that cover the same risk.

Keep three similar words apart, because they get mixed up in planning:

  • An audit exception is a test result from the CPA firm.
  • A finding is your internal record of a confirmed gap that needs an owner, remediation, and verified closure.
  • A policy exception is an approved, time-bound departure from a policy, such as a legacy system that cannot use SSO. If you approve one before the control is tested and the system description covers it, the CPA firm may treat it differently from an unplanned miss. Confirm that with the firm.

The policy exception guide covers the third one in detail.

When do exceptions change the opinion?

The CPA firm decides, so do not treat any list as a formula. In practice, the questions it weighs look like these:

  1. How many sampled items failed, and did the firm expand the sample after the first failure?
  2. How important is the control to the criterion it supports?
  3. Does another control cover the same risk? For example, a quarterly access review that caught the late removals.
  4. Did the problem spread across several controls or systems?
  5. Could a report reader be misled without a modification?

A single late access removal backed by a quarterly review that caught it is a very different story from a change management control that had no approvals for three months. Ask your CPA firm before fieldwork how it handles sample expansion, compensating controls, and exceptions that management found and fixed during the period.

Where exceptions come from

The exceptions you can prevent tend to share a cause: a control that existed on paper but did not run on time or did not leave proof. Some examples:

Area What the CPA firm finds Prevention
Access removal Departed workers kept access past the deadline. A departure checklist with timed steps. See the offboarding guide.
Access reviews A quarterly review was skipped or had no record of decisions. Scheduled reviews that save the user list and each decision. See access review evidence.
Change management Production changes merged without the required approval. Branch protection that matches the written control. See change management evidence.
Training New hires finished security training after the policy deadline. A hire checklist with the training due date.
Vendor review A critical vendor had no review in the period. A dated vendor review cadence. See the vendor review checklist.
Backups No restore test was done, or the test left no evidence. A scheduled restore test with saved output. See backup restore evidence.
Populations The list of changes, hires, or departures did not reconcile. Export complete audit populations with the query and count.

Notice the pattern. Each row is a control that has a deadline or a schedule. The CC4.1 criterion in the AICPA Trust Services Criteria asks the company to run ongoing or separate evaluations to check that the components of internal control are present and functioning. Your own control testing is how you see a late removal in month two instead of in the report.

What to do when the CPA firm reports an exception

You will usually hear about exceptions during fieldwork, before the report is drafted. Use that time well.

1. Check the facts

Ask for the exact sampled items and the evidence the firm relied on. Compare them with your records. If the firm missed evidence you have, such as an approval stored in a different system, give it the source record. Do not create new evidence after the fact or edit dates. If the records support the exception, accept it.

2. Find the root cause

“Human error” is not a root cause. Ask why the control let the miss happen. Was the step manual with no reminder? Did the deadline fall on a holiday? Did a contractor sit outside the HR system that triggers offboarding? The answer decides whether the fix is a process change, an automation, or a scope change.

3. Fix it and record when the fix took effect

A fix after the period cannot change what happened during the period. It can stop the next period from repeating it. Record the fix date and the first evidence that the fixed control ran, because the next report will test from that point forward.

4. Write the management response

Keep it short and factual:

Management response: Two of 25 departures in the period had access removed
4 and 9 days after the end date. Both workers were contractors whose end dates
were not entered in the HR system, so the offboarding checklist did not start.
On 2026-10-12 management added contractors to the HR system and linked their
end dates to the offboarding checklist. A review of all departures since then
found no late removals. No unauthorized access occurred, based on a review of
login logs for both accounts after their end dates.

Every sentence in that response needs a record behind it, including the claim about login logs. Leave out any claim you cannot prove. Ask your CPA firm where the response will appear and whether its opinion covers it.

5. Prepare for customer questions

Security teams that review your report will read the exceptions. Have a short summary ready with the same facts as the management response and the current state of the fix. If the report is older than a few months, a bridge letter may cover the gap until the next report, and it should mention material changes truthfully.

Can a startup get a bad opinion on its first report?

Yes, but you can lower the risk a lot before the CPA firm starts:

  • Run a readiness pass against the audit readiness checklist before the Type 1 date or the start of the Type 2 period.
  • Start the Type 2 observation period only when your controls are running and leaving evidence. A late start beats a period full of exceptions.
  • Test your own controls on a small sample in the first month of the period.
  • Ask the CPA firm whether you can move the Type 2 period start if readiness shows controls are not yet running, and what happens once testing has begun. The auditor selection guide covers questions to ask before you sign.

A Type 1 report tests design as of one date, so it does not list operating exceptions over a period. It can still carry a modified opinion if the description is wrong or a control is not suitably designed. The Type 1 vs Type 2 guide explains the choice.

Track exceptions and findings in FileGRC

FileGRC is a Git-native GRC workspace for SOC 2 work. JSON holds structured records, Markdown holds long-form work, and Git supplies the change history. The audit and each finding are separate records. Each finding points at the audit, and filegrc references lists them from the audit side, so you can trace an exception from the report to the fix and see whether the fix held.

An Audit record stores the opinion once the report is issued, with unmodified, qualified, adverse, disclaimer, or not-issued as the allowed values. It also holds the opinion date, the issued report stored as evidence (reportEvidenceId), and a link to the management response document.

Keep the CPA firm’s tests, samples, and results in the audit’s Markdown. For each confirmed exception that needs its own owner, due date, and verified closure, open a finding whose sourceResourceId is the audit ID and whose controlIds names the control that slipped. FileGRC’s exception record is for approved policy exceptions, so do not use it for an exception the CPA firm reports. Each scaffold command prints a template. Save it in a private temporary folder outside the repository, and run the next steps in the same shell session:

mutation_dir=$(umask 077; mktemp -d)
npx filegrc scaffold finding --title "Late access removal for two departures" > "$mutation_dir/finding.json"

Fill in the severity, description, owners, and due date. Set sourceResourceId to the audit ID and add a controlIds array with the control that slipped. Then preview the change. The preview lists any missing field or unknown ID, so fix the file and run it again until it passes:

npx filegrc preview-mutation "$mutation_dir/finding.json" --json

Then create the finding, remove the temporary folder, and check the workspace:

npx filegrc create "$mutation_dir/finding.json" --json && rm -rf "$mutation_dir"
npx filegrc validate
npx filegrc references audit-soc-2-type-2-2026

The references command lists every record that points at the audit, including the findings you opened against it. Use your own audit ID.

If you also want the auditor’s test as its own record, create a control test with its controlId, testKinds, performedBy: "service-auditor", the auditId, sampleSize, and exceptionCount, and point the finding at it instead. Then run references on the control test ID to see its findings. A finished test needs a complete status, an outcome, externalTester with the CPA firm’s name, evidenceIds, coverage, completedOn, reviewerIds, and reviewedOn. The preview reports missing fields, so run it again after each fix until it passes. Do not leave a test from an issued report as planned.

The finding stays open until someone verifies the fix. A closed finding needs resolvedOn, verifiedByIds, and verifiedOn, so the record shows who checked the fix and when, not only who made it. Put the root cause and the timeline in the finding’s Markdown, and link the evidence that the fixed control ran.

FileGRC writes the files but does not commit them. Review the diff before you commit. When you plan the next period, npx filegrc next-audit-cycle takes the prior audit ID and the new period dates, and --preview shows the result before anything is written.

An agent with access to the repository can read the audit and its findings together, then list which findings are past due, which have no evidence that the fix ran, and which controls had exceptions in the last report. It can also draft the management response from the finding records for a person to check and approve.

FileGRC does not test controls, decide whether an exception is material, or predict the opinion. Your systems and people operate the controls. An independent CPA firm performs the SOC 2 examination, evaluates the exceptions, and issues the opinion.

Open source · MIT

Run your SOC 2 program as files in Git.

Keep policies, controls, work, and evidence indexes in a repository your team and agents can inspect. Add optional FileGRC Autopilot email and Slack reminders to keep work moving.

Frequently asked questions

Can you fail a SOC 2 audit?

A SOC 2 report has no pass or fail grade. The CPA firm issues an opinion. An unmodified opinion is the clean result. A qualified opinion means a material problem affects part of the subject matter. An adverse opinion means the problem is material and pervasive. A disclaimer means the firm could not get enough evidence to form an opinion. Customers read the opinion and the listed exceptions together, so a report with a modified opinion can still cost you a deal even though nobody stamps it failed.

What is an exception in a SOC 2 report?

An exception, also called a deviation, is a case where the CPA firm's test found a control did not work as described. For example, the firm samples 25 departures and finds two where access was removed after the policy deadline. In a Type 2 report, the section that describes tests of controls lists the test, the sample, and the exceptions found.

Does one exception cause a qualified opinion?

Not automatically. The CPA firm judges whether the exceptions, taken together, mean a control did not achieve what the criteria need. It weighs things such as how many items failed, how important the control is, whether other controls cover the same risk, and whether the problem appears across several areas. Ask your CPA firm how it evaluates exceptions before fieldwork starts.

What is a management response to a SOC 2 exception?

It is management's written explanation of what happened, why, and what was done to fix it. Many reports include these responses next to the exceptions. Ask your CPA firm where and how they will appear and whether the opinion covers them. Write facts you can prove: dates, root cause, the fix, and when the fix took effect. A response cannot erase an exception from the period.

Can a startup fail a Type 1 report?

A Type 1 report can still carry a modified opinion, for example when the system description is not fairly presented or a control is not suitably designed as of the report date. It does not test operating effectiveness over a period, so it will not list operating exceptions the way a Type 2 report does. A readiness review before the Type 1 date finds most design gaps early.

Does FileGRC prevent SOC 2 exceptions?

No. Your systems and people operate the controls, and the CPA firm tests them and decides on the opinion. FileGRC is a Git-native GRC workspace for SOC 2 work. It records the audit, the opinion, findings from the report, owners, due dates, and verified closure, with JSON for structured records, Markdown for long-form work, and Git for change history.