Skip to content

Net sentiment

The percentage of a brand's AI mentions that are positive minus the percentage that are negative, on a scale from −100 to 100.

Being mentioned isn't the same as being recommended. An answer that calls you “expensive, with a steep learning curve” still counts as a mention. Net sentiment tells you whether engines are selling you or warning people off.

How Citeplot calculates it

In Citeplot, net sentiment is % positive mentions minus % negative mentions, from −100 to 100. Each mention is scored from −1 to 1. Scores of 0.25 or higher count as positive, −0.25 or lower count as negative, and everything in between is neutral. If half your mentions are positive and a tenth are negative, net sentiment is +40.

Scoring uses an LLM judge when one is configured. Without one, Citeplot uses a transparent lexicon tuned to how engines describe products, and labels the result as estimated. The details are in our methodology.

Fixing negative sentiment

  • Read the full answers. Find the exact phrases engines repeat.
  • Trace them to sources. Negative framing often comes from a review, a forum thread or an outdated comparison.
  • Fix the facts you control: pricing pages, changelogs, docs, comparison pages.
  • Address real weaknesses in public, so there's a newer, credible page for engines to find.

Bring better evidence to your next pitch.

Start a 14-day trial. Explore clearly labelled sample data, then build the evidence for your clients. No card required.