Support staff should not need a customer's password to diagnose a problem. A support workflow needs its own identity, permissions, and history so the team can explain what was viewed or changed.
Separate ordinary customer access from exceptional support access. Otherwise, an admin convenience can become an invisible route across every organisation's private records.
Define support tasks before granting powers
List the tasks support actually performs: inspect a failed job, confirm an entitlement, retry delivery, correct an approved field, or help an owner regain access. Decide which need metadata only and which require viewing private content.
Use separate permissions for diagnosis and repair. A staff member who can see an event reference does not automatically need the power to grant paid access, delete a study, or export all responses. Name an escalation owner for sensitive actions.
Review a fictional paid-access support case
Suppose a customer reports that a purchase succeeded but the study remains locked. A diagnostic view can show the account, study identifier, billing reference, entitlement state, and last reconciliation result without revealing participant answers.
If a repair is needed, require the support actor to select the affected study and a reason. The system should record the staff identity, customer organisation, case reference, previous state, new state, and time. A later reviewer should be able to tell whether access changed through normal billing or a support intervention.
Test that a diagnostic-only role cannot invoke the repair endpoint directly. Test that a support grant for one study does not become permanent access to every study in the account. Test a revoked support role using an old session.
Make exceptional access visible and bounded
If viewing the customer's interface is necessary, define the scope, duration, approval, and customer communication under the organisation's policy. Keep the staff actor distinct in the audit trail even if the interface resembles the customer's view.
Avoid a global customer selector that silently converts every request into an unrestricted tenant context. The server needs to know both the real support actor and the target organisation. That distinction helps enforce limits and preserve meaningful history.
The tradeoff is diagnostic speed versus exposure. More context can shorten a support investigation but broaden access unnecessarily. Start with metadata and add controlled escalation for content or repair. A small team can implement this simply while still preserving actor identity and scope.
The StackFlight product account describes private review and paid study access. This support dashboard is a proposed design, not a description of its internal tools.
Questions about support administration
Should support use customer accounts?
Use a dedicated support identity and workflow where possible. Shared customer credentials obscure responsibility and make revocation harder to reason about.
Does every repair need two people?
Not necessarily. Choose approval rules by consequence and policy. Even a single-person repair should remain scoped, justified, and auditable.
What should a review test first?
Attempt a customer-impacting action as a diagnostic-only staff user, then verify the audit record for an authorised repair. Bring the result to a support-access review.