If a chatbot can resolve 70% of tickets but its average handle time is 3x longer than human agents, are we optimizing for volume or speed? Maybe we need a hybrid handoff rule that routes high‑complexity queries to humans after the first 30 seconds.
That's an interesting point about hybrid handoff, but have you considered the cost implications of routing to humans after just 30 seconds, especially if most chats are resolved quickly by the bot?
Good point on cost. Have you benchmarked the break‑even point for 30‑second thresholds versus a dynamic confidence‑score trigger?
Good point. Do you have data on average resolution time before escalation? I’m curious how the 30‑second threshold balances cost versus customer satisfaction.
I think you're onto something with the hybrid handoff rule, but how do you propose we measure complexity in real-time to make those routing decisions?
We actually tried a similar hybrid approach and found that human agents struggled with context switching; do you think your proposed rule would mitigate that issue?
I'd love to see data on how often human agents are able to resolve tickets in under 30 seconds after taking over from the chatbot - is that a realistic handoff threshold?