Dealing with a very strange bug here with the Queue Performance dashboard metrics in our Genesys Cloud EU (Frankfurt) environment. The ‘Abandoned Calls’ count for the primary support queue is significantly higher than the sum of ‘Terminated by Caller’ events in the Architect flow execution logs for the last 24 hours. Is there a known delay in the Performance view data ingestion, or does the dashboard include calls that disconnected before reaching the flow routing logic?
Have you tried correlating the timestamp of the flow termination event with the queue entry time using the purecloud-architecture-node SDK? This discrepancy often stems from how abandoned calls are defined in the performance dashboard versus how they are logged in Architect. The dashboard counts a call as abandoned if it disconnects before an agent accepts it, regardless of whether the flow itself processed the disconnection event. If the flow logic is slow or has a long wait queue, the call might be marked as abandoned by the queue system before the flow node even registers the termination.
The performance data relies on a batch-processed pipeline that can have a 15-30 minute latency, as seen in other WFM metric discrepancies. However, the specific gap you are seeing is likely due to calls dropping in the initial routing logic before the “Terminated by Caller” node is reached. To verify this, you can query the /api/v2/analytics/queues/summary endpoint and filter for abandoned events, then cross-reference with the flow execution logs. If the flow logs show fewer terminations, it confirms that calls are dropping out before the flow logic can capture them.
A common fix is to add a “Disconnect” node at the very start of the flow to capture all incoming disconnections, even those that occur before any routing logic. This ensures that every dropped call is logged in the flow execution history, allowing for a more accurate comparison with the dashboard metrics. Additionally, ensure that your OAuth tokens have the analytics:queueview scope to access the detailed breakdown of abandoned calls in the API. This approach provides a more granular view of where calls are dropping off in the journey.
The documentation actually says abandonment counts trigger when a caller disconnects before agent acceptance, not necessarily when the flow logs a termination event. This explains the gap if agents are accepting calls after the initial timeout window.
- Queue entry vs. acceptance timestamps
- Flow wait step configurations
- Abandonment definition in WFM metrics