AI sentiment

AI sentiment is a score for how favorably AI answers describe your brand. Our methodology scores how the answers that name you describe you on a scale from 0 to 100, averaged across those answers, where higher is more positive. The idea comes from sentiment analysis, which Google Cloud's documentation describes as identifying "the prevailing emotional opinion within the text," applied here to what each answer says about your brand.

Why it matters

Being named is only part of the result. An answer can name you and then call you "expensive for small teams" or "missing key integrations", and the buyer reads that description next to your name. When sentiment drops, look first at the pages the answer cites: an old review, a forum thread or a comparison written before a feature shipped can carry a negative description into the answer.

How to read it

• Average only over answers that name you. Scoring the others as zero would turn a visibility problem into a sentiment problem.
• Split by prompt group and engine. A healthy average can hide one topic where answers repeat a complaint.
• Read the words behind the number. Find the sentence that pulls the score down and the page the answer cites for it.

Example

A project management tool averages 70 across the answers that name it, but 41 on prompts about resource planning. Those answers cite a comparison page, written before the tool shipped resource planning, that says it lacks the feature. The team asks the page's author for a correction and states the facts on its own site. If no cited page supports a negative claim, it may be a hallucination, and the fix is to make the true facts easy to find and consistent everywhere.