Predicted CSAT
Predicted CSAT
A predicted 1–5 satisfaction score with reasons for every AI and human conversation, no survey needed.
Most customers never answer a satisfaction survey. TR-AI predicts how satisfied each customer is likely to be — for replies sent by TR-AI and by your team — from the question and the reply: was the problem solved, was the answer specific and accurate, did the tone fit, did the customer have to do extra work?
- New conversations are scored every 15 minutes, eight at a time in one small AI call, up to 400 per day.
- The report shows the average, the 1–5 distribution, AI versus human, and the conversations likely to be unhappy (1–2) with the reasons.
- When your help desk reports real CSAT, it is compared with the prediction (how often they are within one point).
Tips & tricks
- Read the “likely unhappy” list weekly — it is the fastest way to find answers that technically resolved a ticket but left the customer frustrated.
- Compare AI and human averages per topic before widening automation for that topic.
- Use “Score new conversations now” after a change to see its effect without waiting.