SOC 2 Trust Services Criteria: The 5 Domains and How to Choose

The SOC 2 Trust Services Criteria are the five categories a CPA firm uses to evaluate how your organization handles client data: Security, Availability, Processing Integrity, Confidentiality, and Privacy. Security is the only category required in every SOC 2 report. You choose which of the other four to include based on what your service actually does and what your clients expect to see. The American Institute of Certified Public Accountants (AICPA) publishes the criteria in TSP Section 100, and they sit on top of the Committee of Sponsoring Organizations (COSO) internal controls framework.

The modular design is the point. A cloud storage platform might need Security, Availability, and Confidentiality but skip Privacy because it never touches personal information directly. A healthcare SaaS vendor handling patient records would likely include all five. Picking the right combination keeps the audit focused on controls that actually matter for your service instead of testing things unrelated to what you deliver.

Security: The Common Criteria

Security is the foundation of every SOC 2 report and the category you cannot opt out of. The AICPA calls these controls the Common Criteria because they apply universally, and they are organized into nine groups known as the CC series.

  • CC1 covers the control environment: how leadership sets the tone, assigns responsibilities, and holds people accountable.
  • CC2 covers communication and information: how security policies, expectations, and relevant information move inside and outside the organization.
  • CC3 covers risk assessment: how you identify, evaluate, and manage threats, including risks introduced by technology or operational changes.
  • CC4 covers monitoring: how you track whether controls actually work, through dashboards, alerts, audits, and periodic reviews.
  • CC5 covers control activities: the specific mechanisms that enforce your policies, such as approval workflows, separation of duties, and audit trails.
  • CC6 covers logical and physical access: passwords, role-based access, encryption, badges, and physical locks.
  • CC7 covers system operations: how you detect, report, and resolve incidents, plus vulnerability management and response planning.
  • CC8 covers change management: how you update systems and applications without introducing new vulnerabilities.
  • CC9 covers risk mitigation: how you address identified risks through vendor management, business continuity planning, and ongoing security reviews.

Because these nine groups map directly to the COSO framework, an auditor is essentially checking whether your organization treats information security with the rigor financial auditors expect for accounting controls. In practice, that means reviewing firewalls, intrusion detection systems, multi-factor authentication, user access logs, and administrative permissions. The auditor wants to see that you have identified where your vulnerabilities are and done something concrete about them.

Availability

The Availability criterion evaluates whether your system is up and running when it needs to be, measured against your service level agreements. If you promise 99.9% uptime in a contract, the auditor wants to see the controls that back the promise.

Testing focuses on disaster recovery plans and whether you have actually run them, backup management procedures, environmental protections such as uninterruptible power supplies, and capacity monitoring. Having a disaster recovery document in a folder is not enough. The auditor wants evidence you ran the drill, measured how long recovery took, and fixed whatever broke.

Organizations that host critical applications or provide infrastructure services almost always include Availability, because downtime is the risk their clients worry about most. If nobody is reading your SLA looking for uptime numbers, this criterion may not belong in your report.

Processing Integrity

Processing Integrity asks whether your systems do what they are supposed to do with the data moving through them. When a transaction enters your system, does it come out the other end complete, accurate, timely, and authorized?

Auditors trace data from the point of entry through each processing stage to look for unauthorized modifications, dropped records, and timing failures. Testing typically involves reviewing error logs, reconciliation reports, and system performance metrics across the audit period.

This criterion matters most for companies that perform calculations, aggregations, or transformations on client data. If your system ingests raw financial data and produces reports, or if you process insurance claims, a processing integrity failure can cascade into regulatory problems for your client. If you simply store or transmit data without changing it, Processing Integrity is usually the wrong criterion for your report.

Confidentiality

Confidentiality protects sensitive information that is not personal data: trade secrets, proprietary source code, unreleased product designs, business strategies, merger details. If a client shares information with you under a contractual obligation to keep it confidential, this criterion covers how well you do that.

Controls in this area typically include encryption at rest and in transit, data classification policies that label information by sensitivity level, strict access restrictions so only the people who need the data can reach it, and secure disposal when the data is no longer needed. Auditors also look at non-disclosure agreements with employees and contractors.

Confidentiality matters most for companies that handle intellectual property or work as outsourced development or consulting partners, where a leak means real competitive harm to the client.

Privacy

Privacy overlaps conceptually with Confidentiality but targets a specific kind of data: personally identifiable information (PII) such as Social Security numbers, health records, financial account details, and contact information. The AICPA maintains its own Privacy Criteria, built around several core principles.

  • Notice: you tell people what personal information you collect and how you use it, and you update the notice when your practices change.
  • Choice and consent: you give individuals meaningful options about how their information is collected, used, and shared, and you get explicit consent where required.
  • Collection limitation: you collect only the personal information necessary for your stated purposes, using fair and lawful methods.
  • Use, retention, and disposal: you use personal information only for the purposes you described, keep it no longer than necessary, and dispose of it securely.

Auditors check whether you actually follow these principles in practice, not just whether you have a privacy policy on your website. If you collect PII from end users on behalf of your clients, Privacy belongs in your report. Companies handling only business-to-business data with no personal information can usually rely on Confidentiality instead.

How to Choose the Right Combination

Security is in your report whether you like it or not. The real decision is which of the other four to add.

Start with your client contracts and service level agreements. If you have committed to uptime numbers, Availability belongs in scope. If your service transforms or calculates data your clients depend on for their own reporting or compliance, Processing Integrity belongs in scope. If clients send you confidential business information under NDA, add Confidentiality. If you collect or process personal information about individuals, add Privacy.

The pattern that shows up most often: cloud infrastructure and hosting providers pick Security, Availability, and Confidentiality. SaaS platforms that touch consumer data add Privacy on top of those. Companies running calculation-heavy processing, such as payroll processors or claims platforms, add Processing Integrity. A vendor that only stores encrypted business documents may need nothing beyond Security and Confidentiality.

Adding a criterion you do not need means paying to test controls that have nothing to do with your service. Leaving out a criterion your clients expect means fielding awkward questions during vendor review and possibly redoing the audit with a broader scope. The scoping conversation with your auditor exists specifically to get this right before fieldwork starts.

Subservice Organizations Affect What Gets Tested

If you rely on third-party providers such as a cloud host or an outsourced data center, their controls affect the criteria you pick. Under the carve-out method, the vendor’s controls are excluded from your audit, your report names them and describes what they do, and clients rely on the vendor’s own SOC 2 report for those controls. Under the inclusive method, the vendor’s controls are pulled into your audit and tested alongside yours, which expands scope and cost and requires the vendor’s cooperation. Your report will also list complementary user entity controls (CUECs), the controls your clients are expected to maintain, and complementary subservice organization controls (CSOCs), the controls your vendors are expected to maintain. These lists define where your responsibility ends.

Criteria Selection Is Separate From Report Type

Choosing your criteria is a different decision from choosing between a Type 1 and Type 2 report. A Type 1 evaluates whether your controls are designed correctly at a single point in time and typically costs between $5,000 and $20,000. A Type 2 evaluates whether those same controls operated effectively over a period of three to twelve months and generally costs $20,000 to $50,000. Both report types cover whichever criteria you selected. Most enterprise clients specifically ask for a Type 2, because it proves the controls worked consistently rather than existing on paper on one particular day.

Who Actually Needs a SOC 2 Report

No law requires a SOC 2 report. It is not a government regulation or an operating license. In practice, clients make it a condition of doing business, which turns a voluntary standard into an effective requirement. The organizations most commonly asked for one include cloud service providers, data centers, managed IT companies, payroll processors, healthcare technology vendors, document management firms, and financial services platforms. The common thread is that someone else’s sensitive data lives in your systems, and their security team wants documented assurance before signing the contract.