Branch Performance Exception Reporting
Submitted by Nader Mahmoud Ghloum Abdullah
Pain point
CEM regularly asks for branch performance, staff performance against benchmark service times, and no-show rates. Producing it means manual pulls from large extracts, hand-comparison against benchmarks, and commentary written from scratch each cycle — by the time it's distributed, it's stale. Branch managers receive a table of numbers with no interpretation and have to scan it for their own issues. Nobody checks whether breaches were actually acted on, so the same branch can miss the same benchmark for months unnoticed.
Proposed solution
A Power Automate flow runs after the daily Qmatic database update, joins it to the benchmark table, and computes variance by branch, service, and staff band — working entirely on aggregates, so no customer data enters the flow. If nothing breaches, it logs a clean cycle and stops. Where breaches exist, it groups them by branch and generates a manager-specific narrative via AI Builder: what breached, by how much, whether it's a trend or a one-off, a likely cause, and a recommended check. Each manager receives only their own exceptions as an Adaptive Card requiring acknowledgment and a logged action. Unacknowledged breaches escalate to regional after 48 hours, and a third consecutive breach flags for regional review regardless of acknowledgment. No-show rate breaches are appended to the same card as a second exception class.
Claude surface
Chat, Cowork
Open question
Benchmark definitions need to be agreed with CEM before anything gets built, or the output is arguable from day one. Staff-level performance reporting is politically sensitive, so it's better to lead with branch- and service-level reporting and make staff-level detail opt-in for managers.