Quick question about the Predictive Routing model behavior in our EU-West BYOC instance.
The Architect flow logs confirm that agent skills are correctly assigned and available for the specific queue, yet the Performance Dashboard shows a significant drop in ‘Matched’ conversations for agents with high historical proficiency scores in that channel. The model appears to be prioritizing agents with lower skill weights, resulting in increased Average Handle Time and a deviation from the expected routing logic. No configuration changes were made to the skill groups or the routing strategy during this period, and the issue persists across multiple queues. Is there a known latency or caching issue with the predictive model updates in this region?