- Quick question about the discrepancy between Architect flow execution times and the Performance dashboard metrics.
- The flow logs indicate a 45-second duration, yet the queue activity view reports 12 seconds for the same interaction batch.
- Is the dashboard aggregating wait time differently than the actual flow runtime?
My usual workaround is to isolating the metric definitions in the analytics configuration, as the discrepancy often stems from how Genesys Cloud calculates ‘handled time’ versus ‘total flow duration’. The Performance dashboard typically aggregates queue wait time and hold time separately from the actual execution logic of the Architect flow. If your flow contains significant pauses, webhooks, or external API calls that are not routed through a queue, the dashboard might not capture that latency in the same way the flow logs do.
The flow log measures the absolute wall-clock time from entry to exit, including any synchronous waits. The dashboard, however, might be filtering out idle time or aggregating based on agent engagement metrics if a queue is involved. To verify this, check the interaction.events table in the analytics API. Look for the timestamp difference between flow.entry and flow.exit events.
{
"select": "timestamp, event_type, interaction.id",
"where": "interaction.id = 'YOUR_INTERACTION_ID'",
"order_by": "timestamp ASC"
}
Flow execution: 45.2s | Queue wait: 0.0s | Agent handle: 12.1s | Total duration: 45.2s
If the agent handle time is around 12 seconds, that matches your dashboard view. The remaining 33 seconds are likely spent in non-queue elements like IVR menus, external integrations, or conditional waits. To align the metrics, ensure that all critical path elements are tagged with appropriate data actions so they appear in the performance reports. Alternatively, use a custom report that sums up the duration of specific flow nodes rather than relying on the standard queue view. This gives a more accurate picture of where the time is being consumed.
The documentation actually says… queue metrics exclude external API latency. My JMeter tests show this gap clearly when webhook calls exceed 2 seconds. The flow timer keeps running, but the dashboard stops counting.
- WebSocket connection timeouts
- Architect pause blocks
- External system response times