A SOC 2 risk assessment is the process your own organisation runs to identify, analyse and treat risks to your commitments and system requirements, under the CC3 series of the AICPA Trust Services Criteria. It is not the SOC 2 audit itself, and it is not an assessment of your vendors’ SOC 2 reports. Its working output is a living risk register and treatment record, and your auditor tests that output as part of the examination.
Search results for this query mix two different things: the internal risk assessment your company performs under CC3, and template kits for assessing your vendors against SOC 2. This guide covers the first. If you were looking for a vendor assessment template, that work belongs to a different part of the criteria (CC9, covered near the end), and downloading a vendor questionnaire will not satisfy CC3.
| Criterion | What it requires | What you produce in practice |
|---|---|---|
| CC3.1 | Specify objectives clearly enough to identify risks to them | Your commitments and system requirements written down as assessable objectives |
| CC3.2 | Identify and analyse risks to those objectives across the entity | The risk register: risks, owners, scores, treatments |
| CC3.3 | Consider the potential for fraud | Fraud risk considered explicitly inside the register |
| CC3.4 | Identify and assess changes that could significantly impact internal control | A change-trigger practice and records of reassessments |
Why listen to us: Atoro is an Irish AI governance and cyber compliance consultancy for software companies, the first consultancy in Europe certified against ISO/IEC 42001 (certificate AIMS-AT-120224, issued by A-LIGN under ANAB accreditation). Atoro states it has delivered more than 200 certifications (atoro.io, checked 08-08-2026), and SOC 2 readiness, including the CC3 risk assessment, is core to that work.
What a SOC 2 risk assessment actually is
The SOC 2 risk assessment is the CC3 series of the Trust Services Criteria, not a separate deliverable bolted onto the framework. Our SOC 2 requirements guide covers that positioning and how CC3 connects the criteria to your control design; this guide covers the working method: how to run the assessment, build the register and survive the audit of it.
The framing that matters, and the one most how-to pages miss: risk in SOC 2 means risk to your commitments and system requirements, not generic IT risk. TSP section 100, paragraph .14 of the Trust Services Criteria defines commitments as “declarations made by management to customers regarding the performance of one or more of the entity’s systems”, generally found in written contracts, service level agreements or public statements. System requirements are how the system should function to meet those commitments, along with relevant laws, regulations and industry guidelines. Every risk on your register should threaten one of those two things. If a risk threatens nothing you have promised and nothing the system must do, it does not belong in the SOC 2 assessment, however real it feels.
The criteria document is explicit that these risks are expected. Paragraph .02 states: “As in any system of internal control, an entity faces risks that threaten its ability to meet the trust services criteria”, arising from factors such as “the nature of the entity’s operations; the environment in which it operates; the types of information generated, used, or stored by the entity; the types of commitments made to customers and other third parties; responsibilities entailed in operating and maintaining the entity’s systems and processes; the technologies, connection types, and delivery channels used by the entity.” You address those risks through suitably designed controls. The risk assessment is how you decide which controls those are.
The four CC3 criteria, criterion by criterion
Each criterion below is quoted from the CC3 series of the 2017 Trust Services Criteria (TSP section 100), with the criterion text verified as reproduced verbatim in published SOC 2 reports. CC3 maps to COSO principles 6 to 9 from the 2013 COSO Internal Control – Integrated Framework, to which the Trust Services Criteria are aligned.
CC3.1: Specify your objectives
“The entity specifies objectives with sufficient clarity to enable the identification and assessment of risks relating to objectives.” (COSO Principle 6)
In a SOC 2 engagement, “objectives” means your commitments and system requirements. So CC3.1 asks you to write down what you have promised customers and what the system must do, clearly enough that someone can point at a risk and say which commitment it threatens.
In practice: pull the security, availability and confidentiality promises out of your customer contracts, SLAs, security whitepaper and marketing claims. Restate each as an assessable objective. “We take security seriously” fails CC3.1. “Customer data is encrypted at rest and in transit, and the production service maintains the availability level stated in section 4 of our standard agreement” passes it, because you can identify risks against it.
CC3.2: Identify and analyse the risks
“The entity identifies risks to the achievement of its objectives across the entity and analyzes risks as a basis for determining how the risks should be managed.” (COSO Principle 7)
Two words carry the weight here. “Across the entity” means the assessment is not an engineering exercise: risks from sales commitments, support processes, HR offboarding, finance and legal all belong if they threaten your commitments. “Analyzes” means each risk gets a reasoned evaluation of likelihood and impact, and that analysis drives the treatment decision. The output of CC3.2 is the risk register itself, covered in the next section.
CC3.3: Consider fraud
“The entity considers the potential for fraud in assessing risks to the achievement of objectives.” (COSO Principle 8)
This is the criterion first-time assessments skip, and auditors notice. Considering fraud means asking, explicitly and on the record, where someone could act dishonestly against your commitments: an insider abusing access to customer data, management overriding a control to hit a release date, a misstated report to a customer about an incident, procurement conflicts of interest. You do not need a forensic model. You need fraud scenarios on the register with the same likelihood, impact and treatment discipline as every other risk, and minutes showing the discussion happened.
CC3.4: Assess change
“The entity identifies and assesses changes that could significantly impact the system of internal control.” (COSO Principle 9)
Your control environment was designed for the company as it was. CC3.4 asks what happens to it when the company changes: a new product line, an infrastructure migration, an acquisition, a key engineer leaving, entering a regulated market, a new law that applies to your customers. In practice this means a defined trigger list, an agreement about who decides a change is significant, and a record each time a significant change prompts a register review. A risk assessment that never moves while the company doubles in size fails CC3.4 even if it was excellent on day one.
Building the risk register
The register is the artefact the whole assessment lives in. There is no AICPA-issued template, so the fields below are the set we recommend because they map each risk to a commitment and to the criteria your auditor will test.
| Field | Purpose |
|---|---|
| Risk ID | Stable reference for minutes, treatments and audit evidence |
| Risk description | What could happen, in plain language |
| Commitment or system requirement threatened | The CC3.1 objective this risk attaches to |
| Trust services criteria affected | Which criteria the risk bears on |
| Risk owner | One named person, not a team |
| Likelihood | Per your defined scale |
| Impact | Per your defined scale, against the commitment |
| Inherent score | Risk before controls |
| Existing controls | What you already do about it |
| Residual score | Risk after existing controls |
| Treatment decision | Mitigate, accept, transfer or avoid, with rationale |
| Treatment actions and dates | What will change, who acts, by when |
| Review date and status | When it was last looked at, and whether it is live |
Worked logic for one row, as an illustration: your standard agreement commits to a defined availability level. A risk reads “single production database failure causes extended outage”. It threatens that availability commitment, touches the security common criteria and any availability criteria in scope, has an owner in engineering, scores high inherent likelihood and high inherent impact, and your replication and backup controls bring the residual score down to a level the business accepts. The treatment decision, and why, is recorded. That linkage, from contract clause to risk to control to residual acceptance, is what CC3.1 and CC3.2 look like when they are working.
Scoring: inherent, residual, and why the scales are yours to define
Score every risk twice. Inherent risk is the likelihood and impact before your controls. Residual risk is what remains after them. The gap between the two numbers is, in effect, your argument that your controls work, which is why auditors care that both numbers exist and that the reasoning between them is visible.
On scales: the Trust Services Criteria do not prescribe a scoring scale. A 3×3 matrix, a 5×5 matrix and a 1 to 10 rating are all workable, and several ranking pages present one scale as if it were the required standard; it is not. What matters is discipline: written definitions for each level (what “likely” means, what “high impact” means in terms of your commitments), the same scale applied to every risk including fraud, and a recorded rationale where judgement was exercised. An honest, consistent simple scale outperforms an elaborate quantitative model that no one can explain to an auditor. For a first SOC 2, keep it simple.
Cadence and keeping the register alive
Be clear about what is and is not a rule here. The criteria require that you identify, analyse and reassess risk; they do not state a frequency. The market practice you will see quoted, “at least annually plus on significant change”, is practitioner convention, not an AICPA requirement, and the same applies to informal staleness expectations such as auditors viewing a register untouched for twelve months as a finding risk. Treat these as sensible defaults: run a full refresh annually, reassess whenever a CC3.4 trigger fires, and agree recency expectations with your auditor before the examination window rather than discovering them during it.
Evidence expectations follow from the cadence. Keep the current register, prior versions or change history, minutes of the review meetings showing who attended, treatment tracking with dates, and the change assessments triggered under CC3.4. In a Type 2 examination the auditor will want these spread across the observation period, so a register that only shows activity in the fortnight before the audit reads as exactly that.
What the auditor tests: Type 1 vs Type 2
The distinction comes from TSP section 100, paragraph .11. A type 1 SOC 2 report “does not contain an opinion on the operating effectiveness of controls nor a detailed description of tests of controls performed by the service auditor and the results of those tests”; it addresses design. A type 2 engagement covers “the suitability of design and operating effectiveness of controls … throughout a specified period”, and the report includes the tests performed and their results.
For CC3, that difference plays out as follows in practice. In a Type 1, expect the auditor to examine whether the risk assessment process is suitably designed: a documented process, a register with the structure described above, fraud and change visibly within scope, risks linked to commitments. In a Type 2, expect the auditor to test whether it operated: reviews happened during the period, treatments were actioned and tracked, CC3.4 changes triggered real reassessments, and residual decisions were made by the people the process says should make them. A beautiful register created three weeks before a Type 2 window closes does not demonstrate operating effectiveness across the period.
The vendor-risk handoff to CC9
Vendors appear twice in SOC 2, and confusing the two appearances is where the SERP’s mixed intent causes real damage. Inside your CC3 assessment, vendors are a risk source: a critical supplier failing can threaten your availability commitment, so that risk goes on your register like any other. What you then do about it, selecting and developing mitigation activities for vendors and business partners, is CC9, the risk mitigation series. TSP 100 paragraph .05 defines the CC9 criteria as covering “how the entity identifies, selects and develops risk mitigation activities arising from potential business disruptions and the use of vendors and business partners”, and the AICPA’s own site confirms CC9.2 addresses vendor management in a SOC 2 engagement.
So: assessing a vendor’s SOC 2 report, with questionnaires or review templates, is CC9 mitigation work. Your CC3 risk assessment is the upstream process that tells you which vendors matter enough to assess. Keep the two artefacts separate and cross-referenced.
Where SOC 2 risk assessments go wrong
These are the failure patterns we see in practice:
- Template theatre. The register is downloaded, filled once and never reopened. It satisfies nothing in a Type 2 and usually fails CC3.4 immediately.
- The security-team silo. Engineering runs the assessment alone, so fraud, contractual and operational risks never surface. CC3.2 says “across the entity”; involve finance, legal, HR, support and sales leadership in at least the identification session.
- Scoring without definitions. Numbers without written level definitions cannot be defended in an audit and drift between reviewers.
- Skipping fraud and change. CC3.3 and CC3.4 are the two criteria most often absent from first-pass registers, and auditors ask about both specifically.
- Wrong artefact. The team downloads a vendor assessment template and believes CC3 is handled. It is not; that is CC9 work.
- Staleness before the window. The register’s last activity predates the observation period. Agree recency expectations with your auditor early.
Frequently asked questions
Is SOC 2 a risk assessment?
No. SOC 2 is an independent examination of your controls against the AICPA Trust Services Criteria, performed by a CPA firm under the AICPA attestation standards. The risk assessment is one thing your organisation does inside that framework, under the CC3 criteria, to decide which risks threaten your commitments and how to treat them. The auditor then tests your risk assessment as part of the examination.
Is a SOC 2 risk assessment the same as assessing a vendor’s SOC 2 report?
No, and search results often blur the two. Your SOC 2 risk assessment is your own internal process under CC3, covering risks to your commitments and system requirements. Reviewing a vendor’s SOC 2 report or sending a vendor questionnaire is third-party risk work, which sits under CC9 as risk mitigation. Both matter; they are different artefacts owned by different criteria.
How often should a SOC 2 risk assessment be performed?
The Trust Services Criteria require the assessment but do not state a frequency. The common market practice is a full refresh at least annually plus a reassessment whenever a significant change occurs, such as a new product, migration, acquisition or regulatory change. Treat that cadence as practitioner convention rather than an AICPA rule, and confirm your auditor’s recency expectations before the examination window.
Is there an official SOC 2 risk assessment template or scoring scale?
No. The AICPA publishes the criteria and points of focus, not templates, registers or scoring scales, and points of focus themselves are guidance rather than requirements. You choose the register format and the scale, whether a 3×3 matrix or a 1 to 10 rating. What auditors look for is internal consistency: defined levels, applied uniformly, with recorded rationale.
What is the difference between inherent and residual risk?
Inherent risk is the likelihood and impact of a risk before your controls are applied; residual risk is what remains after them. SOC 2 practice expects both scores on the register, because the movement from inherent to residual is your documented argument that your controls reduce the risk. A register showing only one score hides that reasoning and invites audit questions.
Who should take part in the SOC 2 risk assessment?
CC3.2 requires risk identification “across the entity”, so it cannot be only the security or engineering team. Bring in whoever owns your customer commitments, typically sales or commercial leadership, plus finance, legal, HR and operations for fraud and process risks. The security team can facilitate and own the register, but the attendance record should show a cross-functional assessment.
What does the auditor test for CC3 in a Type 2 audit?
A Type 2 report includes an opinion on operating effectiveness across a specified period, so the auditor tests that the process ran, not just that it exists. Expect requests for the register and its history, review meeting minutes, treatment tracking with dates, and evidence that significant changes triggered reassessment under CC3.4. A Type 1, by contrast, examines design only and carries no operating effectiveness opinion.
Do we really have to assess fraud risk for SOC 2?
Yes. CC3.3 states that “the entity considers the potential for fraud in assessing risks to the achievement of objectives”, and it is a standalone criterion, not an optional extra. In practice this means recording fraud scenarios on the register, such as insider data misuse or management override of controls, with the same scoring and treatment discipline as other risks. Skipping it is one of the most common first-audit gaps.
If you would prefer to build the register and the CC3.1 to CC3.4 process with a practitioner who has sat through the audit side of the table, Atoro’s SOC 2 readiness work covers exactly this, from commitment mapping through to the evidence trail your auditor will test.
Sources
- AICPA, “2017 Trust Services Criteria for Security, Availability, Processing Integrity, Confidentiality, and Privacy” (TSP section 100; current edition with revised points of focus 2022), paragraphs .02, .05, .11 and .14; document text accessed via full-text mirror of the AICPA publication at leaglobal.com, 19-08-2026; current edition listing at aicpa-cima.com (free account access).
- CC3.1 to CC3.4 criterion text, verified as reproduced verbatim in published SOC 2 reports: metr.org 2025 SOC 2 Type 1 report PDF and Profisee 2024 Type 2 SOC 2 report PDF, accessed 19-08-2026.
- AICPA, “Maintaining high standards for SOC engagements” (professional insights article confirming CC9.2 vendor management), aicpa-cima.com, accessed 19-08-2026.
- AICPA SOC 2 guide product page (October 2022 edition, context for the Type 1 / Type 2 distinction and updated risk assessment guidance), aicpa-cima.com, accessed 19-08-2026.
- Atoro supplied facts: atoro.io, checked 08-08-2026.