CSAT and NPS are the two scores that your users give the bot themselves. CSAT (Customer Satisfaction Score) evaluates a specific answer. NPS (Net Promoter Score) evaluates the entire conversation.
CSAT shows you which answers to correct first. NPS tells you whether the overall experience with the bot satisfies your customers.
CSAT: a vote on an answer
Under a bot answer, the user reads “Are you satisfied with this answer?” and responds with a thumbs-up or thumbs-down. They can then leave a comment or skip.
How CSAT is calculated
CSAT is the share of thumbs-up votes among all votes, converted to a score out of 10.
CSAT = 10 × thumbs-up votes ÷ (thumbs-up votes + thumbs-down votes)
Example: 24 thumbs-up votes and 25 thumbs-down votes give 10 × 24 ÷ 49 = 4.9 out of 10.
Analytics colors the score to help you read it: orange below 7, green from 7 onward, dark green from 9 onward.
Only votes count. An answer displayed without a vote is not included in the calculation.
A comment does not change the score: it explains it.
Enable CSAT
By default, CSAT is offered below all Louis AI answers, meaning the answers that the AI writes based on your knowledge base. For answers written in your flows, enable it step by step:
In the Flow Builder, click an Answer or User Input.
Open the Advanced.
Under Satisfaction rating, enable Enable CSAT survey.
The question then appears below the step's message. On a User Input step, it lets users rate an exchange handled by the AI in the middle of a flow.
If your Use Case is built in the Advanced Builder, click the heart-shaped icon at the top right of the answer block you want to rate.
NPS: a recommendation score
At the end of a conversation in the widget, the user answers “How likely are you to recommend our chatbot to a friend or colleague?” with a score from 1 to 10.
How NPS is calculated
NPS is the share of promoters minus the share of detractors. Each score places the user in a group:
Score | Group | Effect on NPS |
|---|---|---|
9 or 10 | Promoters | Raises the score |
7 or 8 | Passives | Does not count |
6 or less | Detractors | Lowers the score |
Example: out of 100 scores, 38 promoters, 29 passives, and 33 detractors give 38% − 33% = +5. NPS ranges from −100 (all detractors) to +100 (all promoters).
Analytics displays a negative NPS in red, in orange an NPS of 0 to 49, in green an NPS of 50 and above.
Enable NPS
NPS is available only in your website widget, not on WhatsApp or other channels. It is configured in each widget:
Open Integrate → Channels and choose your widget channel.
Open the NPS.
Enable NPS survey is on.
If needed, check Add a follow-up question for your customer: an open-ended question appears after the rating.
If needed, edit the Thank you message.
Keep the proposed Rating question: this is the standard NPS wording, the same across all companies that measure it.
Where to view CSAT and NPS
Open Analytics → Sentiment. The tab offers three views.
Dashboard, Satisfaction section: the CSAT Global card shows the rating and the number of upvotes and downvotes. CSAT Vote Rate is the share of question displays that received a vote: 49 votes out of 314 displays make 15.6%. Comment Rate is the share of votes accompanied by a comment.
CSAT Report: the rating by source, meaning the origin of the response (Louis AI or a Use Case), and, on the right, the rated messages with user comments. Click a conversation to read it in full.
NPS: the score, the Promoters / Passives / Detractors breakdown, the number of ratings (Ratings), and the change over time (NPS over time).
How to read these two ratings
First, look at the number of votes. In general, only a small minority of users vote. Out of 49 votes, a single vote shifts CSAT by 0.2 points.
Those who vote are not a neutral sample. They are often the very satisfied and the very dissatisfied.
Read the comments for a poorly rated response before changing it: they tell you what was missing.
What CSAT and NPS do not measure
The experience of those who did not vote: this is the role of CX Score, a rating out of 10 calculated by AI across all conversations. A high CSAT with a low CX Score often means that only happy customers vote.
The cause of dissatisfaction: a poor rating may target a rule or a price, not the bot. Comments often make this clear.
Delete a comment
If a customer has left personal data in a comment, the GDPR (the European regulation on personal data) may require you to delete it. Your technical teams can delete a CSAT comment with the Data Deletion API: see Export & Delete Customer Data. The comment then appears as CSAT comment deleted.
What’s next
Sentiment report: the entire Sentiment tab, including the topics that most dissatisfy your customers (Dissatisfaction Drivers).
How to improve your CX Score: what to fix first.