Can anyone clarify the calculation methodology for the Occupancy metric when comparing the Performance Dashboard against raw Architect flow data?
We are observing a significant divergence in reported agent utilization for our primary support queue. The Performance Dashboard indicates an average occupancy of 72% for the last billing cycle, which aligns with our WFM forecasts. However, when we extract the actual talk time and after-call work durations directly from the Architect conversation logs, the calculated occupancy rises to approximately 79%.
This variance is impacting our capacity planning models. The environment is running the standard EU-West configuration, and all flows utilize the default state transitions. We have verified that no manual ACW overrides are being applied by agents, and the flow logic correctly captures the end of interaction timestamp.
Is the Performance Dashboard applying a specific smoothing algorithm or excluding certain conversation states (such as brief transfers or dropped calls) that are still present in the raw flow logs? We need to understand if the dashboard metric is adjusted for ‘effective’ working time versus total logged time to reconcile these reports.