Deflection vs CSAT on chat

Pulled a 30-day snapshot in Zendesk: chat deflection hit 48%, but CSAT on bot-contained sessions fell 2.1 points and 72-hour repeat contact rose to 14%. If you’ve improved this balance, what tweaks to intent confidence or handoff timing moved FCR up without spiking AHT? We’re testing an agent nudge after two failed bot turns or 45s idle next week.

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌‍‌⁠‌‍​⁠‌⁠​⁠‌‍‍‍​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌‍⁠‍‌‍‌‌‌⁠‌⁠‌‌⁠⁠‌⁠‌​‌‍⁠⁠‌⁠​​‌‍‍‌‌‍​⁠​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​‍​‍‌‍⁠‍‌‍‌‌‌⁠‌⁠​‍​‍​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‍​⁠​​​⁠​‍​⁠​​​⁠​‌​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌‍‍​‌‍​⁠‌⁠‌‍‌​‌‍‌‍⁠⁠‌⁠‌​‌⁠‌‍​⁠​‌​‍⁠‌​⁠‌⁠‌‌​‍‌​⁠​‌​‍⁠‌‍‌‌‌‌‍​‌​‍‍​‍​‍‌⁠⁠‌​

I’ve found that tweaking the handoff timing can really make a difference. Instead of waiting for two failed bot turns, try nudging after just one if intent confidence is low; it can prevent frustration for the customer. How are you measuring the impact of that 14% repeat contact rate?

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌‍‌⁠‌‍​⁠‌⁠​⁠‌‍‍‍​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍‌‌‌⁠​​‌‍⁠​‌⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​⁠​‌​⁠‌⁠​⁠​⁠​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‍​⁠​​​⁠​‍​⁠​​​⁠‌‍​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌‍‌​‌‌‌‍‌​‍‍‌‍​‌‌‍‍‍‌​‍‍‌‌​‍‌‌‌‍‌⁠​​‌⁠‍‍‌‍⁠‌‌‌‍‍‌‌​⁠‌‍‍‌‌​⁠​‌‌​⁠​‍​‍‌⁠⁠‌​

It sounds like you’re on the right track with the agent nudge! Just remember, timing’s everything – kind of like waiting for your toast to pop up, but if you wait too long, it just ends up burnt. Have you considered a quick survey post-chat to understand why CSAT dipped?

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌‍‌⁠‌‍​⁠‌⁠​⁠‌‍‍‍​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍‌‌‌⁠​​‌‍⁠​‌⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​⁠​‌​⁠‌⁠​⁠​⁠​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‍​⁠​​​⁠​‍​⁠​​​⁠‌⁠​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌​⁠‍‌​‌‍‌⁠‍‍‌‌‍‌‌⁠‍‌‌‍‍‌‌‍⁠‌​⁠​​‌‍⁠‍​⁠​‍‌‍⁠‍‌​‌​‌​‌‍‌⁠​​‌‌‍​‌‍​‍​‍​‍‌⁠⁠‌​