Does anyone understand why the AI bot deflection rates displayed in the Performance dashboard do not match the actual conversation outcomes recorded in Architect flow analytics? We are observing a significant variance in our EU-West Genesys Cloud instance, specifically within the standard voice queues configured for automated customer service. The dashboard reports a 45% deflection rate for the primary self-service flow, while the Architect flow logs indicate that only 30% of interactions actually terminated successfully without agent handoff. This discrepancy becomes more pronounced during peak hours in the Europe/Paris timezone, which complicates our ability to report accurate efficiency metrics to stakeholders.
The environment consists of Genesys Cloud CX with the default AI Bot integration, version 2024.1, utilizing standard predictive routing settings. We have configured the flow to mark conversations as deflected when the bot successfully resolves the intent and triggers a disconnect action. However, the Performance dashboard appears to count any interaction that does not result in an agent transfer as a deflection, including cases where the user abandons the call after receiving an initial automated response. This misalignment suggests that the dashboard metric may be interpreting ‘no transfer’ as ‘deflection,’ rather than requiring a successful resolution flag from the flow.
We have verified that the Architect flow is correctly setting the disposition code to ‘Deflected’ upon successful bot resolution. The issue persists across multiple queues, indicating a systemic reporting behavior rather than a configuration error in specific flows. The conversation detail views confirm that agents are not involved in the majority of these calls, yet the dashboard metrics do not reflect the true success rate of the bot’s resolution capabilities. This affects our key performance indicators for automated service efficiency.
Is there a known limitation in how the Performance dashboard calculates AI bot deflection rates compared to Architect flow analytics? We need to reconcile these figures to ensure accurate reporting on bot effectiveness and cost savings. Any insights into whether this is a expected behavior or a bug in the current release would be appreciated. We are looking for a method to align the dashboard metrics with the actual flow outcomes to provide reliable data for performance reviews.