What is Customer Satisfaction Score (CSAT)?
Customer Satisfaction Score: A survey metric, usually the percentage of customers who rate an interaction 4 or 5 out of 5, used to judge support quality after a conversation.
CSAT measures how satisfied customers are with a specific interaction. After a conversation closes, the customer is asked to rate it, often on a 1 to 5 scale, and CSAT is the share of positive ratings (typically 4s and 5s) out of all responses. It is simple, immediate, and tied to individual conversations, which makes it useful for comparing agents, channels, and topics.
Its weaknesses are low response rates and bias: very happy and very angry customers respond more than everyone else. When comparing AI-handled and human-handled conversations, make sure both are surveyed the same way, and read the comments, not just the score. Some teams now also use models to estimate satisfaction from conversation transcripts, which covers every conversation but needs checking against real survey data.
Why it matters when you build or buy AI
CSAT on AI conversations is your early warning that automation is costing goodwill. Track it separately for AI-resolved, AI-escalated, and human-only conversations. If AI CSAT is well below human CSAT on certain topics, those topics should be routed to people until the AI improves.
Example
CSAT on AI-handled shipping questions matches the human team, but AI-handled return requests score much lower, so returns are routed to humans while the return flow is fixed.
Related terms
More in Support & Ops
Book 20 minutes, or send a brief
A former CTO replies. You leave with a written plan and a fixed estimate.
Prefer LinkedIn? Connect with Harish.