The Performance dashboard indicates a 40% answer rate for our primary outbound campaign, yet the Architect flow logs show a significantly higher volume of answered interactions during the same CET window.
Does anyone know why the dialer metrics would diverge from the flow execution data when using a standard predictive dialing strategy?
I think the discrepancy often stems from how the system timestamps the “answered” state versus when the interaction is fully handed off to the flow. In predictive dialing, the dialer counts an answer as soon as the line goes off-hook. However, Architect flow logs might register the interaction later, once the first media packet is received or the initial IVR prompt plays.
For legal discovery purposes, we always align our bulk export metadata with the dialer’s internal state, not the flow execution time. This ensures the chain of custody remains intact. The answered_at field in the recording manifest is the source of truth for the dialer metrics. If you are pulling data via the Bulk Export API, check the interaction_start versus answered timestamps in the JSON payload.
Here is how the metadata typically looks in the export manifest:
{
"interaction_id": "12345678-abcd-1234-abcd-1234567890ab",
"channel": "voice",
"timestamp": {
"start_time": "2023-10-27T10:00:00Z",
"answered_time": "2023-10-27T10:00:02Z",
"flow_entry_time": "2023-10-27T10:00:04Z"
},
"metadata": {
"campaign_id": "outbound_primary",
"disposition": "answered"
}
}
The gap between answered_time and flow_entry_time can cause the volume mismatch you are seeing. If agents are disconnected before the flow fully initializes, the dialer counts it as answered, but the flow log might show it as a dropped interaction or simply not log the full duration.
Ensure your S3 bucket policy allows for the retention of these detailed manifests. We often see issues where only the audio file is exported, losing the critical timestamp metadata needed for reconciliation. Use the include_metadata flag in your bulk export job configuration to preserve this context. This approach has helped us resolve similar audit trail gaps in past legal holds.