For the complete documentation index, see llms.txt. This page is also available as Markdown.

Recipe: Sentiment and CSAT on close

Score sentiment and CSAT on the whole conversation when it closes. This is the recommended sentiment and CSAT setup.

Why this pattern

  • Triggering on When a chat room is closed lets the AI see the entire transcript.

  • AI Conditions (Advanced) returns a single label; you Branch by AI output to set fixed CSAT and sentiment per label.

  • An Else branch can fall back to AI CSAT Score and AI Sentiment Score.

Steps

  1. Go to Workflow and click New workflow.

  2. Name it, for example Sentiment and CSAT on close.

  3. Trigger: When a chat room is closed.

  4. Condition: AI Conditions (Advanced). Write a prompt that classifies the transcript into one label and replies with only that label (for example very_satisfied, satisfied, neutral, unsatisfied, very_unsatisfied). Use Test on chat history against a few real chats before you save.

  5. Action: choose Branch by AI output. Add labels matching your prompt. Under each label set Set CSAT Score and Set Sentiment Score accordingly. Add an Else branch that runs AI CSAT Score and AI Sentiment Score as a fallback.

  6. Save and enable the workflow.

Example branch mapping

Use labels that match your prompt exactly. Map each to the matching Set CSAT Score and Set Sentiment Score values (CSAT: Very satisfied to Very unsatisfied; Sentiment: Positive / Neutral / Negative).

Re-score historical cases

  1. Open the CX Log.

  2. Filter and select the cases you want.

  3. Use the bulk Run workflow action with a manual (custom) workflow that runs the same Advanced prompt and branching. See Manual Workflows Shortcut (Example 2: re-score sentiment).

Viewing scores

In the CX Log, use the Sentiment score and CSAT score columns and filters. You can also edit scores inline (subject to permission). Summary analytics include CSAT and sentiment on the Analytics Dashboard.