Skip to main content

Sentiment

The Sentiment tab tells you, at a glance, how customers feel in their own words. Each free-text comment is read by AI and classified into one of three categories:

  • Positive — the customer is happy or complimentary
  • Neutral — factual or mixed, with no strong feeling either way
  • Negative — the customer is unhappy, frustrated, or critical

What you see​

When you open the tab for a survey you get:

  • A headline Sentiment score and a count of scored comments.
  • Four metric cards: the overall score, and the percentage of comments that are Positive, Neutral, and Negative.
  • A Sentiment trend chart showing how the score has moved month by month, once there is enough history to plot.

You can narrow the figures by channel using the filter bar (for example, to compare email feedback against IVR feedback).

Scoring comments​

Most comments are scored automatically as responses come in, so the tab is usually ready to read. If you have just imported a batch of responses, or you want to be sure everything is up to date, use the Analyse with AI button.

  1. Open the Sentiment tab for your survey.
  2. Select a channel in the filter bar if you want to focus on one.
  3. Click Analyse with AI.
  4. The page reports how many comments were newly scored — for example "Scored 12 verbatims." — or "Nothing new to score." if everything was already done.
note

Analysing only scores comments that do not yet have a sentiment. Running it repeatedly is safe and will never re-score or change comments that are already classified.

Reading the score​

The sentiment score summarises the balance of positive and negative feeling across all scored comments. A higher score means more positive sentiment overall. Use it as a direction-of-travel indicator rather than an exact measure — the trend chart is usually more revealing than any single number.

tip

A falling sentiment score is a prompt to dig into why. Switch to the Word cloud to see which words are spiking, or the Topics tab to see which themes are driving the change.

When the score is hidden​

If a survey has too few responses to report safely, the tab shows "Below the anonymity threshold" and the score appears as a dash. This protects individual respondents from being identifiable. Collect more responses and the score will appear.