A security incident during the SOC 2 observation period: what happens to the report
Updated
An incident does not by itself cost you the report or change the opinion. TCSA, a consultancy that prepares companies for SOC 2, writes that an incident "does not automatically prevent a SOC 2 report or force a qualified opinion", and that one your monitoring detected, escalated and contained "is evidence for those controls, not against them". What matters is whether a control failed or was bypassed, whether you still met your commitments to customers, and whether the records survive. The CPA firm that signs the report decides. Save the records this week, before logs rotate.
What the three sources say
Three pages were fetched for this guide: a consultancy and two accounting firms. They agree on the main point and cover different parts of the question.
| Source | Who they are | What it covers |
|---|---|---|
| TCSA (Surendra Pal Singh and Saundhi Chauhan; last reviewed August 2026) | Consultancy that prepares companies for the examination. It says the opinion always comes from an independent CPA firm | The opinion, the disclosure criterion, timing within the period, and what to preserve |
| CohnReznick (Remi Franklin and Mica Katz, 19 December 2025) | Accounting firm | The questions that decide the effect on the report, and the reporting options when the opinion is affected |
| Linford & Co (Jaclyn Finney, partner; published 23 May 2018, updated 13 October 2021) | CPA firm that performs SOC 2 audits | Which incidents management describes in the report, and at what level of detail. It does not discuss the opinion |
Sources: (TCSA) (CohnReznick) (Linford & Co). Two limits. TCSA sells readiness work, so its page is detailed where its services are. And the AICPA's own description criteria were not read for this guide: where this page names a criterion, it is on a source's word.
What decides the outcome
CohnReznick frames it as questions the company answers before the auditor does (CohnReznick):
- Did the controls see it? "Were there SOC 2 controls in place that did not identify, detect and escalate the security incident or event?"
- Did someone go around them? "Were the existing controls bypassed?"
- Did routine controls stop during the response? Its example: "Is there a control to perform daily backups that is no longer running due to the outage?" It also notes that people busy with recovery may drop their regular control tasks.
- Is the evidence still there? Ransomware and similar attacks can destroy the documentation that shows controls ran, "which would cause exceptions within the report".
- Did you still meet your commitments? The company should conclude whether the failure affected its ability to meet its "service commitments and system requirements". CohnReznick says that conclusion "impacts the service auditor's SOC 2 opinion and disclosure requirements".
On Availability, CohnReznick is specific: if that category is in scope and the service commitments were not met for the length of an outage, that "could result in a qualified opinion". Whether Availability belongs in your report at all is covered in SOC 2 scope.
TCSA lays the same ground out as five cases. This is one consultancy's reading, summarized here in this page's words (TCSA):
| What happened | Where TCSA says it lands |
|---|---|
| The controls detected, escalated and contained the incident as described | The opinion is unmodified. The incident is described in the report only if it meets the disclosure test below |
| A control failed for part of the period, was fixed, and the replacement ran long enough to test | An exception for the failure window, with the opinion commonly unmodified. TCSA calls that a call made on the evidence |
| A control failed and nothing else caught it: the incident ran undetected, or the response went around the written process | A qualified opinion that names the criteria affected |
| The system description says or implies controls that were not being performed | A qualified or adverse opinion, because the description itself is wrong |
| The incident destroyed the evidence needed to test controls | A disclaimer of opinion, or a qualification for scope |
The difference between an exception and a qualified opinion, and what a reader of your report makes of each, is in SOC 2 exceptions and qualified opinions. TCSA also ties the incident criteria together: it says CC7.2 through CC7.5 "presuppose that incidents occur". A period with an incident in it is a case the criteria were written for.
When in the period it happened
A Type II report covers a window, so the date of the incident changes what can be tested. TCSA's point, in this page's words: an incident early in the period, fixed and then tested for months, reads very differently from the same incident in the last week with a fix nobody has tested (TCSA). This page prints no number of days for how long a replacement control has to run; ask your CPA firm what it needs. How the window is chosen is in the SOC 2 observation period.
Two cases at the edges, both from TCSA:
- Found after the period ended. TCSA says it depends on when the incident happened, not when you found it. If it happened inside the period, it is new information about that period and the auditor reassesses the affected controls. If it happened after the period end but before the report is signed, the auditor asks about it as a later event.
- A Type I report. There is no operating period to test, but TCSA says the disclosure test still applies as of the report date, and an incident can show that a control the description presents was not in place on that date.
Whether it is described in the report
Disclosure is a separate question from the opinion. Linford & Co dates the rule: before the guidance that "officially went into effect December 15, 2018", companies did not have to describe major incidents in the report. Since then, in Linford's words, management includes information about incidents that "(1) occurred as a result of a failure in the design or operating effectiveness of one or multiple controls or (2) upon occurrence resulted in the company not being able to meet service commitments and system requirements" (Linford & Co).
What goes in, per Linford: the nature of the incident, its timing, and its consequence for the company and its users. It lists six questions management can use to decide whether an incident is reportable. In this page's words:
- Did it come from a control that was badly designed or did not operate as intended?
- Did it keep the company from meeting the service commitments and system requirements it gave users?
- Does a law or regulation force public disclosure?
- Does it affect the company's financial stability or long-term operations, enough to need a financial statement disclosure?
- Did it bring a legal or regulatory sanction?
- Has it caused, or will it cause, a major contract to be canceled?
Linford's rule of thumb: "If the answer to any of these considerations is yes, then it is likely that the incident needs to be reported within the upcoming SOC 2 report." It also says "not all incidents are major", and that the description "should be presented at a high level", partly so it does not hand an attacker details.
The sources name the rule differently. Linford gives no criterion number. TCSA names it as description criterion DC4. CohnReznick refers to the same AICPA document as "DC 200" and lists three factors of its own that inform the judgment: "One or more controls were deemed ineffective", "Significant changes were made as part of the company's remediation", and a public disclosure such as a report to a regulator (CohnReznick). Where the description sits in the report is covered in the system description template.
If the opinion is affected: two options
CohnReznick names two routes when the service auditor decides the event affects the opinion: "extend the reporting period to include remediated controls or issue two SOC 2 reports". In the second route the first report includes the incident, with a qualified opinion if the auditor finds one necessary, and a separate report for a shorter period follows it (CohnReznick). Both cost time and audit fees, and neither is yours to pick alone: the auditor has to agree. No price or duration for either route is printed here because no fetched source gives one.
What to save this week
TCSA lists seven things to do while the response is still running, on the grounds that logs rotate, alerts get edited and tickets get re-graded before anyone thinks about fieldwork (TCSA). In this page's words:
- Export the alert that fired, with the detection rule as it stood and the record of who was paged, before anyone tunes the rule.
- Extend log retention on the affected sources (sign-in, API gateway, CI, storage access) before the default rotation deletes them.
- Record severity changes as changes: who changed the rating, when and why.
- Keep one ticket open until the root cause is written and attached, instead of closing at containment and reopening later.
- Snapshot the configuration before the fix, so there is a before and an after.
- Save the pager and on-call records the same week. They back up the times in the ticket.
- Write the timeline from system records within the week, with the timestamps and where each came from.
Policyseed sells a kit, so read the text below as one example of a written rule. It is rendered from the policy templates for the Northwind Cloud Inc sample intake, not typed by hand, so it changes when a template does. If your company adopted these policies, the items are also what the auditor will hold the incident record against.
The incident record (Incident Response Policy, 4.4)
Every incident has a record in the incident log with a unique identifier, timestamped timeline, decisions, participants, affected systems and data, evidence and closure criteria, retained for seven years under the Data Retention and Disposal Policy.
Severity changes (Incident Response Policy, 4.2)
Every report is triaged and assigned a Section 2 severity within the acknowledgment target for that level; severity may be raised by any responder and lowered only by the Incident Commander with a written rationale.
Evidence before remediation (Incident Response Policy, 4.5)
Evidence, including logs in Datadog, cloud audit trails, snapshots and access records, is preserved before any remediation that could destroy it; evidence that may be needed in a dispute is kept in a restricted location with a chain-of-custody note.
How evidence is collected (Incident Response Policy, 5.5)
Evidence collection. Before eradication, responders snapshot affected volumes, export logs and cloud audit trails covering at least 30 days before the first indicator, capture attacker activity and preserve related messages and tickets in a restricted, access-logged location; large artifacts are hashed.
The review afterward (Incident Response Policy, 4.11)
Every SEV1 and SEV2 incident receives a blameless post-incident review within five business days of closure that produces a root cause, impact statement and corrective actions with owners and due dates, tracked to closure and reported to Executive Management.
Three things to check against your own incident. The review has a deadline counted from closure, so the date the ticket closed and the date of the review will both be read. Lowering a severity needs a written reason from the Incident Commander, which is the same record TCSA asks for. And evidence is captured before the fix, so a host that was rebuilt first is a departure from the policy that should be written down, not left for the auditor to find. The incident response plan template has the operational steps, and the tabletop exercise page lists what to keep for the auditor from an exercise.
Notification is not a SOC 2 rule
Who you must tell, and how fast, comes from your contracts and from law, not from the SOC 2 criteria. TCSA lists customer contracts and legal duties beside the report disclosure and says of them: "different audience, different trigger, different clock" (TCSA). This page gives no legal deadline; that is a question for counsel. What SOC 2 does test is the promise in your own policy:
Customer notification (Incident Response Policy, 4.8)
Affected customers are notified without undue delay, within any contractual timeframe, and in every case within 72 hours of Northwind Cloud Inc confirming their data was affected, stating what happened, what Northwind Cloud Inc has done and what the customer should do.
The notification assessment (Incident Response Policy, 5.7)
Notification. Within 24 hours of confirming that customer or personal data was affected, the Security Owner prepares a notification assessment listing who must be told and by when; Executive Management approves it and the messages, counsel reviews regulatory notices, and copies of every notice are attached to the incident record.
The hours in item 4.8 are the policy's own number, so if customer data was affected, the time between confirming it and the first notice is evidence either way. Keep a copy of every notice with the incident record, as item 5.7 says.
When the incident was at a vendor
CohnReznick counts these in: events "either at your organization, vendor or subservice organization may disrupt services provided to customers including the performance of SOC 2 control activities" (CohnReznick). TCSA adds that management should consider disclosing a known incident at a subservice organization whether the report carves that vendor out or includes it, and that the auditor will look at how you learned of the incident, how quickly, and what you did (TCSA). The generated policy treats a vendor incident as your incident:
Vendor incidents (Incident Response Policy, 4.10)
Incidents at vendors, including Supabase, Stripe, Datadog and Slack, that affect Northwind Cloud Inc data or Northwind are handled under this policy, and vendor contracts must require notification to Northwind Cloud Inc within 72 hours under the Vendor and Third-Party Risk Management Policy.
So open your own record when the vendor's notice arrives, note when it arrived, and write down what you checked. The two ways a report treats a vendor are in carve-out vs inclusive method, and the ongoing review in SOC 2 vendor management.
A short list for this week
This list is the page's own suggestion, built from the items above, not a list from a source.
- Finish the response first. Containment and customers come before the audit file.
- Export before you edit. The alert, the logs, the pager record and the configuration, in that order of how fast they disappear.
- Name the control in the root cause. Write which control failed or was missing, and the date the fix took effect.
- Check what else stopped. List the routine tasks that were skipped during the response (reviews, backups, scans) and record each in the policy exception register if it broke a written rule.
- Call the CPA firm with the timeline and the root cause, and ask how they will treat it.
- Answer customers with dates. If a deal is open, say what happened and where the fix stands. The trust page generator and sharing your SOC 2 report cover what to publish and what to send on request.
Frequently asked questions
- Does a security incident mean we fail SOC 2?
- No, by the sources fetched for this page. TCSA, a consultancy that prepares companies for the examination, writes that an incident "does not automatically prevent a SOC 2 report or force a qualified opinion", and that one your own monitoring detected, escalated and contained "is evidence for those controls, not against them". CohnReznick, an accounting firm, puts the weight on whether controls were bypassed or missed the event, and whether you still met your service commitments. The CPA firm that signs the report decides.
- Do we have to disclose the incident in the SOC 2 report?
- Only some incidents. Linford & Co, a CPA firm, says management includes incidents that came from a control that was badly designed or did not operate, or that kept the company from meeting its service commitments and system requirements. The disclosure covers the nature of the incident, its timing and its effect, and Linford says it "should be presented at a high level". TCSA names the criterion as DC4. Whether your incident meets the test is a judgment made with the CPA firm.
- Do we have to tell our auditor about the incident?
- Tell them early. CohnReznick's advice is to "Consider disclosure requirements with your service auditor", and Linford says a company with a reportable incident "should work with appropriate personnel such as their auditor and legal team". TCSA adds that management signs written representations that cover known control deficiencies, and that "the standards are far more tolerant of a bad event than of a concealed one". An incident log that is exported in full at fieldwork will show the incident anyway.
- Does an incident at one of our vendors count?
- It can. CohnReznick writes that events "either at your organization, vendor or subservice organization may disrupt services provided to customers including the performance of SOC 2 control activities". TCSA says management should consider disclosing a known incident at a subservice organization whichever way the report treats that vendor, and that the auditor will look at how you learned of it, how quickly, and what you did. That makes your own vendor monitoring record the thing to have ready.
- What if nothing happened all year?
- Then say so and show that the process ran. On the disclosure side, TCSA writes that where nothing meets the threshold, "management may say so explicitly". On the control side, keep the alerts and reports that were triaged and closed as not incidents, and the record of your tabletop exercise. No fetched source says how an auditor samples an empty incident list, so ask your CPA firm what they will accept.
- Can we move the observation period to leave the incident out?
- TCSA's answer is "It rarely works." Its reasons: the period dates are printed on the front of the report, so a window that was shortened or shifted invites the question, and the description criteria can still call for disclosure of an earlier incident that was not fully fixed during the period. CohnReznick points the other way: when the opinion is affected, one option it names is to extend the period so it includes the remediated controls.
Policyseed provides governance policy templates and AI tailoring. It is not legal advice and not a compliance guarantee. Management adopts the policies; the CPA firm performs the SOC 2 examination.