Quick question about a persistent divergence in reported bot interaction metrics within the EU-West environment. We are utilizing standard Architect flows for an AI-driven self-service tier, routing via a specific IVR menu. The Business Process Dashboard indicates a 78% successful resolution rate for the ‘Account Balance Inquiry’ intent, yet the Conversation Detail Views for the same time window show a significantly higher volume of escalations to human agents.
The specific concern lies in the definition of a ‘successful’ bot interaction. The dashboard appears to count an interaction as successful if the bot provides a response, regardless of whether the user explicitly confirms resolution or hangs up immediately after the initial greeting. In contrast, the Queue Performance View for the associated escalation queue shows a spike in conversations that were transferred but never answered, suggesting the bot may have failed to transfer properly or the user abandoned the call post-bot interaction.
We are not observing any API errors or 5xx responses in the system logs. The Architect flow logic includes a standard ‘Transfer to Agent’ block when confidence scores fall below 0.8. However, the metric gap suggests that the ‘Bot Success’ KPI might be calculating success based on intent detection rather than actual user satisfaction or task completion.
Could you clarify how the Genesys Cloud Performance Dashboard calculates ‘Bot Resolution’ versus ‘Bot Escalation’? Specifically, does the metric account for user hang-ups occurring within 10 seconds of the bot’s final response? We need to align our reporting with actual operational efficiency, and the current dashboard view seems to overstate bot performance by ignoring immediate post-interaction abandonment. Any insight into the underlying calculation logic or a way to adjust the metric definition in the custom dashboard view would be appreciated.