How do I correctly to reconcile the discrepancy between Bot Satisfaction scores and Agent Satisfaction scores when the handoff rate exceeds 40%?
The Performance Dashboard displays a high satisfaction metric for the bot segment, yet the downstream agent metrics show a significant drop in resolution quality for those transferred conversations. The current view does not provide a unified metric that accounts for the quality degradation occurring during the handoff process.
Please advise on how to configure a custom metric or view that accurately reflects the total customer experience across the bot-to-agent transition.
the documentation actually says satisfaction metrics are calculated independently based on the interaction segment, not the entire journey. when you have a high handoff rate like 40%, the bot satisfaction score only reflects the experience before the transfer happens. it does not automatically inherit the negative sentiment from the agent segment if the user was happy with the bot’s initial routing or information gathering. this creates that illusion of high bot performance while downstream quality drops.
from a workforce management perspective, this disconnect often masks real adherence issues. if agents are receiving complex transfers that the bot failed to resolve, their handle times and quality scores will suffer, but the bot metric stays clean. to get a true picture, you need to look at the “resolved by bot” metric alongside the satisfaction scores. if the bot satisfaction is high but resolution is low, it means users are satisfied with the bot’s attitude or speed but not the outcome.
try building a custom report in the analytics module that joins the bot_interaction_id with the subsequent agent_interaction_id. filter for transfers where the bot satisfaction > 4 out of 5 but the agent quality score is < 80. this will highlight the specific intents causing the degradation. also, check if your IVR is forcing a transfer without capturing the user’s intent clearly, which leads to agent frustration. the key is to stop viewing these as separate metrics and start correlating them at the interaction level to see where the handoff friction is actually occurring. this helps in adjusting both the bot scripts and agent training materials for those specific complex scenarios.