Get Transfer Metrics

Whole-window aggregates with prior-period comparison. Pickup/missed are first-queue rates (downstream answers count as missed).

Sorted by missed rate, worst first

Where calls were actually HANDLED (credited teams, most talk time first) — the complement of teams, which is where they were SENT. Empty until the customer's PBX queues are configured.

Intake volume in the window, on its own denominator.

Deliberately NOT a row in the team tables: a lead can be created on a call no team ever touched (the AI handled it end to end), so intake's denominator is larger than the transferred-call table's. It is a second dimension — what KIND of call it was — crossed with the team tables' question of who handled it.

Individual answerers across all teams, most answers first; AI bounce-backs included

Every second of conversation in the window, by who was talking.

The parts are disjoint and sum to total_seconds exactly, so the page can show the total with its sources side by side and the table's grand total can reconcile against it.

Every inbound call the AI answered in the window, by how it ended. The parts sum to total_calls.

Team roster with inbound, outbound and total talk per person, the AI receptionist as a member of every team it took calls for. Most total talk first; the no-team group last.

Column totals for the workload table, plus the AI talk no team row can own (its conversation BEFORE a transfer, and calls it handled alone for callers with no case on a team), so grand_total_seconds equals the talk-time card's total.