No, SOC 2 does not require SSO. For startups, SOC 2 asks you to control who gets access to your systems, change it when roles change, remove it when people leave, and show evidence that you did; your auditor judges whether your controls meet the criteria. SSO is one way to run some of those controls. A documented, logged process is another.
That said, a startup going through SOC 2 often meets SSO anyway: in an auditor's suggestions, in a customer's security questionnaire, or in a contract. Here is what the framework actually asks about access, what evidence looks like, and where SSO earns its price. None of this is legal or audit advice; your auditor decides what satisfies them.
What SOC 2 is, briefly
SOC 2 is a report, not a certificate. An independent CPA firm examines a service organization's controls and reports on them against the AICPA's Trust Services Criteria. The AICPA describes those criteria as control criteria for evaluating and reporting on controls over security, availability, processing integrity, confidentiality or privacy (AICPA). The current version is the 2017 criteria with revised points of focus from 2022.
Two things follow from that. First, you choose the categories in scope. Security, often called the common criteria, is in every SOC 2; the others are added when they matter to your customers. Second, the criteria describe outcomes, not products. They say what your controls should achieve. They do not tell you which tools to buy, and they do not mention SSO.
Reports come in two types. A Type 1 reports on whether your controls are suitably designed at a point in time. A Type 2 also tests whether they operated effectively over a period, commonly several months to a year. Customers usually want the Type 2.
What SOC 2 says about access
Within the security criteria, one group covers logical and physical access, numbered CC6. In general terms, the logical access criteria ask for four things.
Secure access to your systems and data. Decide what needs protecting, and how people prove who they are before they get in.
Register and authorize people before giving them access. Then remove their access when it is no longer authorized, for example when they leave.
Grant, change and remove access based on roles and responsibilities. Keep least privilege, meaning only the access the job needs, and separation of duties in mind. The principle of least privilege in a 20-person company shows what that looks like at small scale.
Restrict privileged access, such as admin rights to production systems.
Each criterion comes with "points of focus", which are illustrations of what a control might consider, not a checklist you must tick. That leaves room for judgment, and that judgment is your auditor's. The full text is a free download from the AICPA once you register.
So the honest answer to "what access controls does SOC 2 need?" is: controls that cover joiners, movers and leavers, keep access matched to roles, and protect admin rights, designed by you, written in your policy, and tested by your auditor. The joiners, movers and leavers process for small teams is a good frame for designing them.
Evidence: onboarding, offboarding, reviews
For a Type 2, the auditor will want proof that the controls ran throughout the period. In practice, auditors often take lists of who joined and who left during the period and pick samples to check. Each part of the access lifecycle needs its own records.
Onboarding
For each new person, keep a request or ticket showing who approved their access and which role it was based on. Then keep evidence that the accounts created match that role: no extra tools, no higher roles. If the role says "engineer" and the person also got admin rights in your billing tool, that is the kind of gap a sample turns up.
Role changes
When someone moves team or takes on a new job, record what changed, including access that was removed, not just added. Movers are where access quietly piles up. The person who moved from support to sales usually keeps the support tool, because nobody asked for it back.
Offboarding
For each leaver, keep the date the person left, from HR or the directory. Keep a record of each system they were removed from, with a timestamp. And be able to show that the removal happened within the window your policy sets. If your policy says one business day, a leaver removed from GitHub after two weeks is an exception the auditor will note.
How do you prove offboarding to an auditor? With records made at the time: a ticket or checklist per leaver, timestamps from each tool's admin log, and a directory showing the account suspended. Screenshots taken months later are weaker. The employee offboarding checklist gives you the steps to record.
Reviews
Run a periodic review of who has access to each in-scope system, with the frequency set by your own policy. Quarterly is a common choice for critical systems, and user access review has a process and a template. Record what the reviewer found and what was changed as a result, and keep a sign-off showing who reviewed, and when.
Write the policy first and keep it short. Access control policy for small businesses has a template. An auditor tests you against what you said you would do, so do not promise what you cannot keep.
Contractors, service accounts and other edge cases
The people an auditor samples are not only employees.
Contractors with access to in-scope systems need the same joiner and leaver records as staff. They are also the people most likely to be missed, because they often leave without an HR exit date to trigger anything. Give each contract an end date and treat that date as the leaving date. Contractor access management covers the practical side.
Shared and service accounts, such as a deploy user or a shared support inbox login, still need an owner and a reason to exist. An account that belongs to nobody is hard to justify in a review.
Founders and early employees often hold admin rights everywhere because they set everything up. That is normal in year one, and exactly what the privileged access criterion will make you look at. Decide who really needs admin, and record why.
Where SSO helps, and where it does not
SSO, single sign-on through an identity provider (what SSO is), helps with parts of this.
It puts authentication in one place. MFA and password rules are set once in the identity provider and apply to every connected system, which is easy to evidence with one configuration screen.
It gives you one switch for logins. Suspending a leaver in the identity provider blocks new logins to every SSO-connected system at once.
And it keeps login logs. The identity provider records who signed in to what, which helps when investigating.
It does not help with the rest. It does not remove accounts: accounts stay in each tool after the person is blocked. If your policy says access is removed, an account that still exists, holding a role, may be read as access that was not removed; that depends on the tool and on your auditor. It does not do reviews, so someone still has to compare each tool's member list against who should be there. It covers only connected tools, so in-scope systems outside SSO need their own controls. And it is expensive for startups, because most tools keep SSO on their Business or Enterprise plans (the SSO tax).
Do you need SCIM for SOC 2? No. SCIM, the standard that lets an identity provider create and remove accounts automatically (what SCIM is), makes removals faster and leaves a clean trail. The criteria ask that access is removed when no longer authorized; they do not ask how. A manual removal done on time and recorded can meet them.
When customers ask for SSO specifically
SOC 2 may not require SSO, but your customers can. It usually shows up in three places.
The first is security questionnaires: "Do employees access production systems through SSO with MFA?" Answer truthfully. "Google Workspace sign-in with enforced 2-Step Verification for our cloud console and code host; other tools behind MFA and a documented offboarding process" is a real answer, and many buyers accept it.
The second is contract clauses. Some enterprise agreements require that staff with access to the customer's data log in through an identity provider. If you sign one, you need real SSO for the systems that clause covers. A process document will not satisfy it.
The third is SSO into your own product. Often "Do you support SSO?" means "Can our people log in to your app through our identity provider?" That is a feature of your product, not a control on your staff, and a separate project from SOC 2.
Read the question carefully before you price an upgrade.
A practical SOC 2 setup for a startup
For a startup of 5 to 50 people heading for its first SOC 2, the setup can be built in a sensible order.
Start with one directory. Google Workspace or Microsoft 365, with every employee and, where possible, every contractor in it. Enforce MFA on the directory, and on every in-scope system that allows it. Google Workspace security has the settings worth turning on.
Then list the in-scope systems. Your cloud provider, code host, production database access and customer-data tools. It is usually a short list, and it decides where the effort goes.
Use company sign-in where it is already included. Many tools offer "Sign in with Google" or Microsoft on your current plan. Buy SSO only where it is required or the risk justifies it, for example your cloud console and code host if a customer contract names them.
Write the access policy next. Roles, who approves access, the removal window, and the review frequency. Keep it to what you will actually do.
Once the period starts, record as you go. Make a record for every joiner, mover and leaver at the time it happens, run the quarterly access reviews of in-scope systems with sign-off, and keep one folder of evidence, organized by control, so the audit is a hand-over rather than a hunt.
Then ask your auditor, early, how they will test access. They decide, and they would rather tell you before the period starts than write up exceptions after.
Where SmashSSO fits
First, plainly: SmashSSO has no SOC 2 report. If you add it to your stack, it is a vendor without one, and your own SOC 2 process may ask you to assess that. It is also not SSO: if a contract or your auditor requires login through an identity provider, buy SSO for those systems.
What it does help with is the access lifecycle. It reads your people from Google Workspace, shows who has an account in each tool and with what role, and flags accounts that match nobody in the directory, which is a useful starting point for a review. It adds people by role through access bundles, and when someone is suspended in Google Workspace it alerts you and shows where they still have access, then removes them. Every change is recorded in its audit log, and any step that would add a paid seat pauses for approval. Changes run through a tool's member API where you have added a token, or through an agent in your own Chrome where you are signed in as an admin; that agent needs the browser open and can fail when a tool changes its screens. There is no CSV export, so for auditor samples you will still pull user lists from each tool.
It costs a flat fee from $199 a month billed annually (pricing). The security page describes how it handles your data.