Predictive Routing skew on BYOC trunks during peak load

Looking for advice on why predictive routing metrics show a 40% drop in answer rate for our 15 BYOC trunks in Asia/Singapore during peak hours. The Architect flow logs indicate successful outbound calls, yet the WFM dashboard reports dropped calls.

  • Verified SIP 200 OK responses on all trunks via packet capture.
  • Checked outbound routing priorities and confirmed no carrier failover loops.

Any insights into metric discrepancies between Architect and WFM?