How can a queue outage become a practical pilot acceptance test?

An environment builder scopes a pilot around a professional services billing team. A customer cannot meet an invoice and agrees to speak with the accounts support specialist. Referral REF-0728 is ready to submit, but the receiving queue returns an unavailable response.

The configured fallback allows the adviser to save a pending request and record a contact preference. It does not allow the adviser to report that the specialist team has accepted the case. The referral changes from ready to pending-delivery, and the customer receives an accurate explanation of that state.

When the queue becomes available, the environment retries delivery using the same request identity. A successful acceptance changes the referral to accepted. Replaying the retry must not create another specialist case.

The buyer asks MirrorFirm to generate the starting records, fallback policy and expected action outcomes for this sequence. The builder supplies the queue adapter and retry mechanism.

Pilot acceptance rests on visible evidence: the pending request survives interruption, the customer message matches its state and recovery produces a single accepted referral. This defines a useful division between generated content and infrastructure behaviour without requiring a full account-management simulator.