A second transfer request can arrive while the first destination is still ringing. Treat it as a new decision with an explicit relationship to the existing attempt, rather than letting both transfers proceed independently.
Transfer requests need attempt identity
A second transfer request changes the desired destination or asks for another handoff after an initial connection. Those situations require different state transitions. Capture whether the original attempt is pending, completed, failed, or already being cancelled.
For a proposed workflow, retain one current caller intent and separate transfer-attempt records. Each attempt has its destination, reason, creation time, and terminal state. Cancelling the first attempt should prevent its late events from changing the second attempt’s outcome.
Do not reset the entire interaction when the destination changes. Keep verified caller details and the original reason for calling, while updating the handoff summary to reflect the latest request. Preserve earlier attempts in the audit trail.
Test a destination change during ringing
A fictional caller initially asks for sales, then says, “I already bought it; I need repairs.” The agent should clarify only what is necessary, cancel the sales offer, and route to the repair workflow. It should not require the caller to repeat contact details already confirmed.
Suppose sales accepts just after cancellation is requested. The application needs an explicit resolution: either reject that stale acceptance or complete a controlled recovery before offering repairs. Merely sending a cancellation command is insufficient evidence that the old offer ended.
| Sequence | Expected outcome |
|---|---|
| New destination before ringing starts | Only the new offer is sent |
| New destination during ringing | Old attempt reaches a terminal state |
| Old acceptance arrives late | No unexpected sales connection |
| Repair destination unavailable | Caller receives a truthful fallback |
Add a case where the caller repeats the same destination rather than changing it. That repetition should usually check or explain the current attempt, not create another ringing leg.
Limit transfer loops without trapping callers
A caller may legitimately reach the wrong department. Allow a further handoff when policy permits, but preserve enough history for the next employee to understand the mistake. Repeated bouncing between the same destinations should trigger a different owner or callback route.
A strict transfer limit controls cost and complexity but can block legitimate resolution. Make the fallback meaningful and visible to staff. The point is to resolve ownership, not to win a counter.
This is a proposed state-management approach. See the voice integration case study for related experience and transfer rejection recovery for the failure branch.
Second transfer FAQ
Should every repeated request create a new attempt?
No. Determine whether the caller is changing intent or asking about the current offer.
Can two destinations ring at once?
Only as an explicitly designed group-routing policy with a single winner and reliable cleanup.
What context should the second employee receive?
Include the current request, confirmed facts, earlier destination, and the reason another handoff became necessary.