Sentiment

How it is computed

Every brand mention is scored individually, from 0 to 100, based on how the surrounding answer frames the brand. The sentiment metric is the average of those per-mention scores over the responses matching the active filter, rounded to a whole number.

sentiment = round(average of the sentiment scores of the brand's mentions)

When the brand is not mentioned at all in the filtered period, sentiment has no value rather than defaulting to a number, since an average over nothing would be misleading.

Reading the scale

The scale is anchored at 50 for a neutral, purely factual mention. As a rule of thumb, the bands read as follows.

Around 90

The brand is strongly endorsed, for example recommended outright as the best choice.

Around 70

The brand is framed positively, with clear strengths attached to it.

Around 50

The brand is mentioned neutrally, listed without praise or criticism.

Around 30

The brand is mildly criticized, for example named with a notable drawback.

Around 10

The brand is strongly criticized or explicitly advised against.

Most brands average in the 50 to 70 range, because most AI answers list options factually rather than argue for them. A stable 55 is not a problem to fix; a slide from 65 toward 45 is worth reading into.

Interpreting movement

Because the metric averages over mentions, it moves both when the framing changes and when the mix of mentions changes. A new prompt whose answers discuss your weaknesses will pull the average down even if every previous answer is unchanged. When sentiment moves, you should filter to the period in question and read the individual responses; each mention carries its own score, so the answers responsible for the shift are easy to find.

Where it appears

Sentiment appears on the project overview alongside visibility, has its own chart over time, and is listed per brand in the brand table, so you may compare how the platforms frame you against how they frame each competitor.