Designing a Provisioning Test
Key Takeaways
Choose sandbox data that distinguishes account creation from updates and revocation.
Test a Create Account calculation by actually creating a new target account in the sandbox.
Include positive, negative and boundary cases with an explicit expected target state.
Capture the request or trigger, provisioning evidence and a native readback for each write test.
Turn the requirement into a measurable statement
A useful provisioning test specifies the identity, source, account, access, starting state and expected change. “The employee receives finance access” leaves open which permissions and which account. A more useful acceptance condition is that an eligible employee without an account on the test finance source receives exactly one new account in the approved group, with the configured work email and no privileged group. Set a separate condition for a worker outside the qualifying population. Establish these expectations before running identity processing or placing requests.
SailPoint recommends testing new provisioning configurations in a sandbox before enabling them in production. Choose a controlled target and seed representative data. Use the actual intended connector type, account model and trigger. A sandbox that uses manual tasks can prove a task-routing path. To test a production direct write, use a test target with the equivalent direct-connect configuration and appropriately scoped credentials. Record which connector and permissions were used so differences in environments are visible.
Build cases around initial conditions
| Test | Before and trigger | Observable result |
|---|---|---|
| Create | Eligible identity, no target account, requested grant approved | One new account, correct generated name and required fields, expected membership |
| Modify | One existing target account, grant another supported entitlement | Correct existing account changes, with unrelated membership and identity preserved |
| Remove | Account holds one entitlement exclusively from the removed role | Entitlement gone and account still present after revocation |
| Deny | Requested access withheld by the planned approval decision | No write for that request and no change to target membership |
| Missing input | Test mapping returns no value for a field the sandbox target requires | Account creation rejected by the configured validation or target. Verify exact error and inspect for partial account changes |
| Wrong account | One person with separate ordinary and privileged accounts, only the ordinary account intended for the grant | Apply the intended account selection path and verify ordinary account changed, privileged account unchanged |
For each row, capture what would count as a failure. If you expect to keep an unrelated group, compare that membership explicitly. If testing a boundary, make the inputs differ only on the condition under test. A role eligibility test might pair one qualifying department and an otherwise identical nonqualifying department. For account selection, compare an unambiguous account with a pair of matches to expose a broad filter. Test only the intended action type so the cause of the observed change remains clear.
Exercise the real calculation
Suppose a change calculates the new account’s display name from given and family names while copying Work Email directly. An identity profile preview can help you check those identity inputs. Test the source’s Create Account calculation by granting access to a sandbox identity with no existing target account. A grant to an existing account exercises an update. The policy change may never run in that test. Choose a second fresh identity with a different name or a test collision where the configured generator must resolve uniqueness. Capture the calculated account values and validate the resulting account attributes from the source itself.
When testing failure recovery, arrange a controlled rejected action such as a missing required input and log the exact operation. Check for a partial account before reusing the fixture. Use a fresh identity or follow the approved cleanup procedure to restore the starting state. Establishing the same baseline matters when comparing before and after a policy change.
Decide what evidence is sufficient
For an automated lifecycle or role test, capture the identity processing input and resulting state or assignment. Identity processing can initiate provisioning, so open the resulting provisioning activity for the actual operation. For access requests, capture the approval outcome and then check execution. If the case uses a service desk or manual task, capture the task, fulfillment and relevant status mapping. For successful changes, compare native target account values and membership to each expected outcome. Mark a case incomplete if the only evidence is a pending request or a control-plane configuration.
See SailPoint’s sandbox guidance, Create Account transform testing, identity processing, and provisioning tracking.
An engineer needs to test a changed Create Account transform. Which sandbox scenario exercises the change?
Preview the identity profile and stop after the identity name appears
Aggregate a target that already holds the test user’s account
Grant access to an eligible test identity with no existing target account and inspect the created attributes
Add an entitlement to an existing account and assume the create transform also ran
The test case requires that a denied request leave an existing target account unchanged. What evidence completes the case?
A denial record and native readback showing unchanged membership on that account
A successful connector Test Connection
A role configuration screenshot with a different test user
A provisioning task marked pending without checking the target
Sections you finish are checked off in the contents.