What is the standard approach to maintain predictive routing scores during a mid-call carrier failover? We manage 15 BYOC trunks in Asia/Singapore and see score resets when Genesys Cloud v2.61 switches carriers mid-session. The Analytics API shows a gap in engagement metrics. Does the routing engine re-evaluate attributes upon SIP re-registration, or is there a context variable we need to preserve in the Architect flow to prevent this degradation in Predictive Routing accuracy?
TL;DR: Context variables don’t survive SIP re-registration.
This looks like a session state issue rather than a routing logic flaw. The platform treats the failover as a new leg, so you need to pass the original interaction ID via SIP headers to link the analytics back to the initial engagement.