How do you calculate CSAT?
A CSAT survey asks one question soon after an interaction, such as “How satisfied were you with the help you received?”, with answers from 1 (very dissatisfied) to 5 (very satisfied). The most common score is the share of satisfied responses: count the 4s and 5s, divide by the total number of responses and multiply by 100.
For example, if 200 customers answer and 150 choose 4 or 5, CSAT is 75%. Some tools report the average rating instead. Microsoft Copilot Studio, for instance, shows AI agent owners a satisfaction score out of 5 from end-of-session surveys, and groups ratings of 1 and 2 as dissatisfied, 3 as neutral, and 4 and 5 as satisfied, according to its documentation checked on 8 October 2026.
Two related surveys measure different things. Net Promoter Score asks how likely someone is to recommend you, and Customer Effort Score asks how easy it was to get what they needed.
How should you use CSAT for an AI agent?
Ask at the end of the conversation, while it’s fresh. The UK government’s Service Manual tells teams to let users say what they think once they’ve finished using a service, and to include an open question about how it could be better.
Read the comments behind low ratings. For an AI agent, a poor rating can mean a wrong answer, a missing answer, a misheard question or a customer who wanted a person, and each needs a different fix.
Watch how many people respond. Customers with strong feelings are more likely to answer, so a small number of responses can swing the score.
Does Kawl Now measure CSAT?
No. Kawl Now doesn’t run a satisfaction survey or calculate a CSAT score. When a call ends, the caller can press “That was helpful”, which is saved on the conversation. There’s no negative option, so helpful presses are a positive signal only: a caller who doesn’t press it may have been satisfied, unhappy or simply finished.
In your dashboard, “Feedback captured” marks any conversation with saved feedback. That can be a helpful press, or a bug report, feature request or other feedback the agent recorded because the caller asked it to, so open the conversation to see which it was.
What you do get is the evidence behind satisfaction: a text transcript of every conversation, with the customer’s words as transcribed and the agent’s replies, plus its topic, status and length. Reading the transcripts shows where the agent didn’t know an answer or misunderstood a question. The Insights tab also groups recurring questions, reported issues and themes from recent calls, with quotes from the transcripts, but it doesn’t score satisfaction.
If you need a CSAT figure, run the survey in your own tools, such as a short form on the page where the widget sits. The widget doesn’t ask callers for an email address, so there’s no address to send a survey to afterwards.
Common questions
What is a good CSAT score?
There’s no universal benchmark, because scores depend on the question, the scale, when you ask and who answers. Compare your score with your own earlier results from the same survey, and read the comments behind low ratings to see what to fix.
What is the difference between CSAT and NPS?
CSAT asks how satisfied someone was, usually with a recent interaction, on a short scale such as 1 to 5. Net Promoter Score asks how likely they are to recommend you on a scale of 0 to 10, and subtracts the percentage of detractors (0 to 6) from the percentage of promoters (9 and 10). CSAT suits a single conversation, while NPS describes the wider relationship.