← All posts
SOC 2 for single-tenant SaaSSOC 2 dedicated customer deploymentSaaS deployment evidence

SOC 2 for Single-Tenant SaaS: Account for Every Deployment

A dedicated customer instance can fit a SaaS SOC 2 scope. Map shared services, list every deployment, and show which controls operated in each environment.

A dedicated customer stack still depends on shared deployment, support, and review paths.
A dedicated customer stack still depends on shared deployment, support, and review paths.

For SOC 2, a single-tenant SaaS team needs to show which service it operates, which customer deployments that service includes, and whether the relevant controls ran across those deployments. A separate stack per customer can make some data boundaries easier to describe. It also creates more places where a release, access rule, backup job, or monitoring setting can differ. Start with a deployment inventory and a map of the shared paths that operate the fleet.

What does single-tenant SaaS change for SOC 2?

The AICPA SOC 2 Description Criteria ask management to describe the system placed in operation to provide its services, including relevant infrastructure, software, people, procedures, and data. The description must not imply that a control operated when it did not. The criteria do not require a shared or dedicated deployment model.

AWS’s SaaS Lens calls dedicated resources a silo model. Even a full customer stack can use shared identity, onboarding, deployment, support, and monitoring. A service with both dedicated and shared components may be a bridge model. Those are useful architecture terms, not additional SOC 2 criteria. The multi-tenant SaaS guide covers isolation checks inside shared paths. Here the main question is whether the team can account for each vendor-operated deployment when it makes a service-wide control claim.

If the customer operates the instance instead, the boundary changes. Use the customer-hosted software guide to separate the vendor’s release and support duties from the customer’s runtime controls.

Draw the shared path and the customer stack

Pick one customer action, such as signing in and starting a job. Trace it from the shared control plane, if any, into that customer’s environment. Then trace the paths that let your staff deploy, grant access, troubleshoot, restore, or export data. Mark who operates each component and where source records live.

Part of the service Question to answer Source to inspect
Customer stack Which account, database, storage, region, software version, and settings serve this customer? Deployment manifest, cloud inventory, configuration
Shared control plane Which identity, CI, release, support, logging, or monitoring systems can affect several stacks? Architecture, access grants, pipeline settings
Cross-stack access Can an operator, service role, or automation reach more than one customer environment? Role policy, approved support case, audit event
Recovery Is backup and restore configured and tested for this deployment type? Backup job, restore test, exception record
Provider and customer Which cloud services and customer actions are needed for the service commitment? Provider terms, customer contract, responsibility map

Use the SaaS scope guide to connect that map to the service customers bought and the commitments you made. Agree on the proposed report description with the independent CPA firm. A statement such as “dedicated infrastructure” says little about a shared release key or support role that can reach every stack.

Make the deployment list a checkable population

Before selecting examples, create a dated list of every vendor-operated customer environment in the proposed scope. Include active, newly provisioned, retired, and exceptional deployments where they fall in the relevant date or period. Record a stable deployment ID, customer reference, start and end dates, region or account, software version, configuration class, owner, and source inventory revision. Keep sensitive customer identifiers in an access-controlled source system when the GRC repository should not retain them.

Reconcile that list against billing or provisioning records, cloud accounts, deployment targets, and monitoring targets. These systems may disagree. An unlisted stack is a gap to investigate, not a row to leave out because it makes the sample harder. The audit population guide shows how to keep the source, cutoff, inclusion rule, and reconciliation conclusion together. Management and its CPA firm decide which population and sampling approach fit the examination.

Trace one release through the fleet

Choose a real product change that should have reached all in-scope stacks. Follow its approved source revision to the build artifact, the release plan, each deployment result, and a final reconciliation against the deployment population. If one customer remains on an older version, record why, who approved the difference, which controls it affects, and when it will be reviewed again.

The same completeness check applies to recurring work. If a monitoring rule is required for every stack, compare configured targets with the deployment population. If the team claims every stack has backup coverage, inspect job status and the restore-test approach for each relevant configuration class. Keep failures and exceptions visible. AWS notes that siloed environments add operational and monitoring complexity, so a single successful deployment or screenshot does not establish fleet-wide operation. The change evidence guide shows the source-to-production chain in more detail.

For a Type 2 examination, retain dated records for the agreed period. Do not project today’s fleet state backward onto a deployment that existed earlier. The AICPA description criteria distinguish a Type 2 examination’s control-operation subject matter from a Type 1 examination’s point-in-time design subject matter. Ask the CPA firm what evidence it needs for the actual scope and period.

FileGRC is a Git-native GRC workspace for SOC 2 work. In the FileGRC workspace, JSON holds structured records, Markdown holds long-form work, and Git supplies the change history. Your team can link the deployment map, control owners, population reconciliation, evidence references, and exceptions so an agent can inspect the whole program and flag missing coverage. Starter records are proposals, not compliance claims. Cloud, CI, identity, backup, and monitoring systems operate the controls and produce source evidence. Your team reviews the facts and answers; an independent CPA firm performs the examination and decides whether the evidence is sufficient.

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 a single-tenant SaaS service get a SOC 2 report?

Yes. A dedicated deployment model does not prevent a SOC 2 examination. Management must describe the service and the system it actually operates, including dedicated customer stacks and relevant shared services. The independent CPA firm agrees on the examination scope and evaluates the controls and evidence.

Does one SOC 2 report cover every dedicated customer instance?

Do not assume it does. Read the issued report's system description, date or period, and any stated boundaries. Before the examination, list the deployments your team operates, their material differences, and the controls that apply. Agree with the CPA firm on how the report will describe and test that population.

Do dedicated customer stacks remove the need for tenant isolation controls?

A separate stack can change the data boundary, but shared identity, deployment, monitoring, backup, and support paths may still affect more than one customer. Test the boundaries that exist in your actual service and keep source evidence for them.

What if one customer runs a different software version or security setting?

Record the difference, the approval behind it, the affected commitments and controls, and the source evidence. A standard control result for other deployments does not prove that the changed customer stack followed the same path. Discuss material variations with the CPA firm before stating what the report covers.

Can FileGRC verify every customer deployment automatically?

No. FileGRC organizes the SOC 2 program as JSON records and Markdown work with Git change history. An agent can compare a deployment inventory with linked control and evidence records when it has access to current sources, but the cloud, CI, identity, and monitoring systems provide operating proof. Your team reviews gaps and the independent CPA firm judges evidence sufficiency.