Noru

How to Write a SOC 2 System Description (DC 200)

What the SOC 2 system description must contain under the AICPA DC 200 description criteria, how to write each section, and the mistakes that cause rework.

Bip Thelin3 min readPublished

The SOC 2 system description is management's own account of the system the report covers, written against the AICPA's DC 200 description criteria, and it is what the auditor's opinion is about. It sets out the services provided, the service commitments and system requirements, the components of the system (infrastructure, software, people, procedures and data), any significant incidents, the criteria and related controls, the complementary user entity and subservice organization controls, and any significant changes during the period. The auditor gives an opinion on whether the description is fairly presented, so it has to match the system as it actually ran.

A SOC 2 report is usually organized into a handful of sections: the auditor's report, management's assertion, the description of the system, and, in a Type 2, the criteria, controls, tests and results. The system description is the longest piece management writes and the one first-time companies most underestimate.

What DC 200 asks for

The AICPA's description criteria for SOC 2, known as DC 200, set out what the description has to disclose. In plain terms:

ElementWhat to write
Types of services providedWhat the service is and who uses it, in terms a customer would recognize
Service commitments and system requirementsWhat you promise customers about security and the other categories in scope, and the requirements the system has to meet to keep those promises
System componentsInfrastructure, software, people, procedures and data that make up the system
Significant incidentsSystem incidents that caused a failure to meet commitments or requirements, and their nature, timing and extent
Criteria and related controlsThe Trust Services Criteria in scope and the controls that address them
Complementary user entity controlsWhat the design assumes customers do themselves
Subservice organizationsWhich vendors perform part of the service, how they are treated, and the controls you assume they operate
Criteria not relevantAny criterion that does not apply to the system, and why
Significant changesChanges to the system during the period, for a Type 2

Drawing the boundary

The boundary decides what is in scope. A clear description names the production environment, the services and data stores inside it, the tools used to build and deploy it, and the people and teams who operate it. Leaving something out because it is untidy, such as an internal admin tool with production access, is the kind of omission an auditor finds in walkthroughs, and it forces a rewrite.

Writing the system components

  • Infrastructure: cloud providers and regions, networks, environments and how they are separated.
  • Software: the application, its main services and the third-party software it depends on.
  • People: the teams that build, operate and govern the system, and their responsibilities.
  • Procedures: how access, changes, incidents, monitoring and vendors are handled, in summary.
  • Data: the kinds of data the system processes and stores, how it flows and how it is protected.

Commitments and requirements

Service commitments are what you have promised customers in contracts, terms, SLAs and public statements, such as encrypting data or notifying them of incidents. System requirements are what the system has to do to keep those promises. They are the reference point for the whole report: the controls exist to meet them, and an incident is significant when it caused a failure to meet them.

Complementary controls and subservice organizations

The description lists what you assume customers do and which vendors perform part of the service, with the controls you assume each one operates. The guide to CUECs and subservice organizations covers carve-out and inclusive treatment in detail.

Mistakes that cause rework

  • Describing the target state instead of what ran during the period.
  • Copying a template that names tools or procedures you do not use.
  • A boundary that leaves out systems with access to production or customer data.
  • Procedures described in more detail than the controls actually support, which the auditor then has to test.
  • Forgetting to record significant changes that happened during the period.

The AICPA's 2027 exposure draft also revises the description criteria alongside the Trust Services Criteria; the article on the 2027 criteria covers what is changing.

Primary sources

Related reading

FAQ

SOC 2 questions answered

Talk to us

Who writes the SOC 2 system description?

Management of the service organization. Auditors and consultants can review drafts, but the description is management's representation, and the auditor gives an independent opinion on whether it is fairly presented.

How long is a typical system description?

Often somewhere between ten and thirty pages, depending on the size of the system and the criteria in scope. Clarity about the boundary and components matters more than length.

Does a Type 1 need a system description too?

Yes. Both report types include the description. A Type 1 describes the system as at a date, and a Type 2 describes it throughout the review period, including significant changes.

Can we reuse last year's description?

As a starting point, yes. Update it for every change during the new period, such as new infrastructure, vendors, teams or incidents, because the auditor checks it against the system as it ran.