Context:
We are operating a Genesys Cloud environment with a significant volume of inbound voice traffic routed through Predictive Routing queues. The primary objective is to maximize first-call resolution while maintaining strict adherence to service level agreements (SLAs) defined in our enterprise contracts. Recently, a discrepancy has emerged between the routing decisions executed by the Architect flow and the metrics reported in the Performance Dashboard.
Specifically, the Architect flow is configured to route interactions based on a custom skill-based predictive model. The flow logic prioritizes agents with a high historical resolution score for specific ticket categories. However, the Queue Activity view indicates that a substantial portion of interactions are being assigned to agents with lower predicted scores, resulting in an unexpected increase in Average Handle Time (AHT) and a decrease in first-call resolution rates. The dashboard shows these interactions as “Routed” but the agent performance metrics suggest the predictive model is not being applied as intended. We have verified that the skills are correctly assigned to the agents and that the queue configuration allows for predictive routing.
Question:
Can anyone clarify why the Performance Dashboard might reflect routing outcomes that contradict the configured Predictive Routing logic in the Architect flow? We are observing that the “Predicted Score” metric, which should drive the routing decision, does not appear to correlate with the actual agent assignments visible in the Conversation Detail view. Is there a known latency or synchronization issue between the routing engine and the dashboard reporting that could cause this mismatch? Alternatively, could there be a configuration nuance in the queue settings that overrides the predictive model under specific load conditions, such as when the queue depth exceeds a certain threshold? We need to understand if this is a reporting artifact or a fundamental issue with the routing execution, as it impacts our ability to validate the effectiveness of our current skill-based routing strategy. Any insights into how to reconcile these two data sources would be greatly appreciated.