Context:
In our Genesys Cloud environment (Europe/Paris region), the Performance dashboard shows significant variance between Bot Conversation duration and subsequent Queue Handle Time. The bot successfully hands off to an agent, yet the handle time metric appears to include the entire bot interaction phase, skewing agent performance reports.
Question:
What is the correct way to configure the Architect flow or dashboard metrics so that Handle Time strictly reflects the agent’s interaction post-handoff, excluding the preceding bot conversation duration?
Have you tried isolating the metric definitions in the dashboard configuration? The default handle time often pulls from the initial interaction start, which includes the bot phase. You need to explicitly filter for the agent-assigned segment. In the dashboard builder, set the metric scope to “Agent Handle Time” rather than “Interaction Duration.” This ensures the timer starts only when the conversation is transferred to the human queue, excluding the automated pre-processing steps.
From a load testing perspective, this discrepancy can also stem from timestamp synchronization issues during high concurrency. If the bot transfer happens within milliseconds of the agent pickup, the API might batch the events. Check your Architect flow for any unnecessary delays in the transfer step. A simple “Wait” action or a specific “Transfer to Queue” node with explicit timeout settings can help separate the bot duration from the agent duration in the analytics pipeline.
Also, verify the data refresh interval in your dashboard. Real-time metrics often lag behind the actual transaction completion during peak loads. If you are seeing this during heavy traffic, the 503 errors mentioned in previous threads might be causing partial data writes. Ensure your query window is large enough to capture the full lifecycle, but small enough to avoid including unrelated historical data. Adjusting the granularity to hourly instead of real-time might provide a clearer picture of the actual handle times.