When Does Business Software Need Single Sign-On?
When does business software need SSO?
Business software needs single sign-on when customers need to manage access through their organisation's identity system and apply consistent account controls. Make it a priority when customers require it for purchasing, when administrators need to control employee sign-in centrally, or when separate application passwords conflict with their access policy. Treat removing access as a separate requirement: SSO alone does not guarantee timely user deactivation.
The useful question is less “Are we large enough for SSO?” and more “Who must control access, and what must happen when that access changes?”
What does single sign-on actually control?
With single sign-on, an application accepts proof of authentication from a trusted identity provider instead of checking a separate application password for that sign-in.
For a product owner, separate the access requirements into three decisions:
- Authentication: how someone proves who they are.
- Authorisation: what that person may do inside the application.
- Account lifecycle: how access is created, changed, suspended and removed.
Specify each separately. For example, require an employee to sign in through the customer's identity provider, give them a purchasing role in your application, and remove their access when their employment ends.
Do not accept “SSO supported” as the complete description of those behaviours. Ask the build partner to explain which system owns each decision.
For the broader context around protecting application data, read data security in modern applications.
Which customer needs justify building it?
Use customer requirements to decide priority rather than choosing an arbitrary company-size threshold.
| Customer situation | Suggested decision | What to establish first |
|---|---|---|
| A customer requires central employee sign-in | Prioritise SSO | Their identity provider and acceptance requirements |
| Administrators want central control over who can enter | Scope SSO with account lifecycle controls | How suspension reaches the application |
| Different customers use different identity systems | Define a compatibility plan | Which connections you will support and test |
| Most users register independently | Validate demand before building | Whether an organisation actually manages their access |
| The request is simply “make login easier” | Clarify the underlying problem | Whether convenience or central control is the goal |
A useful discovery question is: “What must your administrator be able to do without contacting our support team?”
Turn the answer into observable requirements. “Disable an employee and prevent further access” is clearer than “enterprise-ready authentication”.
What does identity-provider compatibility involve?
Common federation standards include SAML and OpenID Connect. Supporting a standard does not, by itself, establish that every customer's configuration will work with your product.
Ask prospective customers which identity provider they use and which connection method their administrators expect.
Before promising compatibility, agree on:
- The information required to configure the connection.
- The identity attributes the application needs.
- Whether users start sign-in from the application, the identity provider, or both.
- How administrators test a connection before enforcing it.
- How configuration changes and expired credentials are handled.
- Who can change the connection and recover from a broken setup.
Make compatibility claims specific. Prefer “tested with your agreed configuration” over an unqualified promise to work with every identity provider.
Include a failed-connection test in acceptance criteria. The customer should see a useful error and a clear recovery route.
How should existing accounts be linked?
Account linking decides whether an incoming identity belongs to an existing application account or requires a new one.
An email address can change or be reassigned, so matching email addresses alone is not sufficient evidence for automatically linking accounts.
Require a stable identity reference scoped to the trusted identity-provider connection. Document what happens when that reference changes or a customer replaces its provider.
For existing users, specify a controlled linking process. Depending on the product, that might require proof of access to both accounts or an authorised administrator's reviewed action.
Resolve these cases before launch:
- An employee already has an account using a work email address.
- A person belongs to multiple customer organisations.
- A returning employee has an old, suspended account.
- A customer changes its email domain.
- Two accounts appear to belong to the same person.
Keep access organisation-specific. Require a deliberate decision before granting membership or merging data; do not make either an automatic consequence of a matching address.
Why does SSO alone not guarantee timely deactivation?
Disabling someone at the identity provider can prevent a new SSO authentication while an existing application session remains usable, depending on the integration and session policy.
That distinction is central to offboarding.
Treat an employee's departure as a checklist covering the application account, active sessions, organisation memberships, alternative login methods and independently issued credentials.
SCIM is a standard used for provisioning and managing user identities between systems. Whether a particular integration suspends accounts or removes access depends on its supported operations and implementation.
Ask for a complete explanation of the removal path:
- What event starts deactivation?
- How does the application receive it?
- Which access paths are disabled?
- What happens to active sessions and personal access tokens?
- How are failures detected and retried?
- How does the administrator confirm completion?
Set an explicit acceptance target for removing access. Do not assume that “central login” answers any of those questions.
What should a worked offboarding test prove?
Consider this illustrative scenario: an employee uses SSO, has an active browser session, and has created a personal access token.
An administrator disables the employee through the agreed identity-management process.
The acceptance test should check that:
- A fresh sign-in is rejected.
- The existing browser session loses access within the agreed period.
- The personal access token stops working according to the agreed policy.
- A password reset cannot restore access through a local login route.
- Other organisation memberships follow their own explicit access rules.
- An authorised administrator can confirm the outcome.
Also test a failed deactivation delivery. Require a visible failure and a recovery process, rather than treating a sent request as proof that access has ended.
For practical guidance on defining these checks, see software testing and quality assurance.
How should emergency access work?
Decide how an authorised administrator will regain control if the identity connection fails.
Specify a narrowly controlled emergency route with strong authentication, named ownership, recorded use and a review after activation. Keep it separate from ordinary employee access.
An alternative local login can bypass an SSO requirement if the application continues to accept it for affected users.
For that reason, define exactly who may use emergency access, what it permits, and how it is disabled or secured again afterwards.
Test recovery before enforcing SSO across a customer organisation. Include the case where the person who configured the connection is unavailable.
What common mistakes should product owners avoid?
Buying a login button instead of an access policy. Require account linking, enforcement, offboarding and recovery decisions alongside the sign-in screen.
Treating successful sign-in as complete verification. Include rejected sign-ins, suspended accounts, broken connections and failed lifecycle updates.
Assuming central sign-in determines application permissions. Document how roles are assigned and who can change them.
Enforcing SSO before testing recovery. Make a successful recovery exercise part of the rollout checklist.
Building without a customer requirement. If users manage their own accounts and no organisation needs central control, investigate other access improvements first.
Keep the initial scope focused on the customers and access rules you can support reliably.
What should founders ask before approving the work?
Ask the build partner to demonstrate the whole journey: configuration, first sign-in, existing-account linking, role assignment, suspension and emergency recovery.
Request clear acceptance criteria for each step. Also establish who handles customer setup, how connection changes are reviewed, and what evidence an administrator receives when access is removed.
Approve SSO as an access-management capability with defined boundaries, rather than a feature label.
Frequently asked questions
Does SSO replace multi-factor authentication?
SSO and multi-factor authentication serve different purposes: SSO establishes a federated sign-in relationship, while multi-factor authentication requires multiple authentication factors.
Specify the authentication assurance your customers require and how the integration will enforce it.
Must every customer use SSO?
Make that a product-policy decision. You can require SSO within an organisation that has enabled it while supporting a different sign-in method for other customers. Define guest and administrator exceptions explicitly.
Should SSO and automated offboarding launch together?
If the customer's requirement includes timely access removal, include a verified removal process in the launch scope. That process may use automated provisioning or an agreed administrative workflow, but it needs an owner, a target and evidence that it works.
*Need to define access requirements for your business software? Start a conversation.*