When users abandon a confusing chatbot or escalate to an agent, conversational outcome metrics reveal where the interaction failed and how much improvement matters. section 3.4.2
These metrics describe outcomes such as appointment completion, abandonment, and escalation. Teams combine them with user feedback and conversation analysis to identify recurring pain points, then prioritize fixes by frequency, implementation cost, and expected benefit. section 3.4.2
PharmaBot appointment booking
Analytics showed frequent escalations and abandoned conversations when PharmaBot confused COVID-19 testing with vaccine appointments. The team prioritized clearer appointment handling because successful booking without agent intervention was the core goal. section 3.4.2
Figure 3.15 Analysis of an increase in escalated conversations within the Schedule Appointment flow, identifying a potential cause (new wording directing users to agents) and providing recommendations to improve
Use the escalation analysis to connect a rising outcome metric with a wording change, likely cause, and recommended improvement. section 3.4.2
