Metrics
Sentiment
How Ranqo labels each AI mention of your brand as positive, neutral or negative, and turns the labels into a 0 to 100 score where 50 is neutral.
Being named is not the same as being recommended. Sentiment measures how AI talks about you when it does name you.
Sentiment: How positively AI talks about you, scored 0 to 100. Neutral mentions count half, so 50 is neutral.
One label per mention
Every answer that names your brand gets one label. A model reads the answer in context, not word by word, and decides:
- Positive: you are recommended, praised, described favourably or ranked highly.
- Neutral: you are mentioned factually, with no clear opinion.
- Negative: you are criticised, described unfavourably or warned against.
Reading in context matters because single words mislead. "Not expensive" is positive, "not the best choice" is negative, and "despite some issues, it's recommended" is positive. If the model's reading cannot be used, Ranqo falls back to a word-list check around your first mention.
An answer that does not name you has no sentiment. It counts against your visibility, never against your sentiment.
The score
The labels are combined into one number from 0 to 100:
Sentiment equals positive mentions + half of neutral mentions divided by labelled mentions, times 100.
- All positive scores 100.
- All neutral scores 50, the middle of the scale.
- All negative scores 0.
- A negative mention costs twice what a neutral one does.
This is the net sentiment measure, the share of positive minus the share of negative, rescaled to run from 0 to 100. It is not the share of positive mentions. Under a plain positive share, a brand that AI mentions only neutrally would score 0 and read as negative, which misstates what AI said.
A worked example
On one engine, AI names you in 46 answers: 25 positive, 21 neutral and none negative.
- Positive counts in full: 25.
- Neutral counts half: 21 × 0.5 = 10.5.
- Sentiment is (25 + 10.5) / 46 × 100 = 77.
A quick reading of the counts gives 25 / 46 = 54%, which is the positive share, not the sentiment score. The difference is the neutral mentions, which count half.
Labels
The Sentiment card shows the score, rounded to a whole number, with a label:
| Score | Label |
|---|---|
| 75 and above | Positive |
| From 55 | Mostly Positive |
| From 45 | Neutral |
| From 25 | Mostly Negative |
| Below 25 | Negative |
When no answer in the range names you, there is nothing to score and the card shows -- rather than a neutral 50.
These labels are the dashboard's. The PDF export, the free AI Visibility Report's email and the hosted report describe the same score with thresholds of their own: a score the dashboard calls Mostly Positive reads Neutral or Positive in the PDF export and the email, and "Mixed tone" in the hosted report. The number is worked out the same way; only the words differ.
Across engines
Like visibility, sentiment is scored on each engine first, then combined with the same engine weights, rescaled over the engines where you have labelled mentions. The positive, neutral and negative counts under the Sentiment card are plain totals across all engines, so the score can differ a little from what the counts alone would give. See Visibility for how the weighting works.
A score of 50 can mean two different things: every mention neutral, or as many positive mentions as negative ones. The counts under the card tell you which.
Where you see it
- Visibility page: the Sentiment card, with the positive, neutral and negative counts beneath it. See Visibility.
- Answers: when you open an answer, the sentences that read as clearly positive or negative about you are highlighted, so you can see what drove the label.
- Competitors page: each competitor's sentiment, scored the same way from the same answers. See Competitors.
- Snapshots: a run's page shows this score. The run list's Sent is an older measure on a different scale. See Snapshots.