What is analyzed
After each run, a language-model pass reads every answer that has text, including answers that name neither you nor a tracked competitor. For each brand the answer discusses, it extracts passages:
Link addresses are removed before analysis, so a URL is never read as an opinion.
Polarity rules
- Positive: the brand is recommended, praised or favorably compared.
- Neutral: the brand is described factually. Facts are neutral.
- Negative: the answer makes an actual unfavorable judgment: it criticizes, discourages or unfavorably compares the brand. When a passage is unclear between neutral and negative, it is classified as neutral.
Aspects
Passages are labeled with one of: Pricing, Ease of Use, Features, Data Quality, Performance, Integrations, Support, Trust & Reputation, Market Position, or General.Answer-level sentiment
Each answer that names your brand also gets one overall label toward your brand: positive, neutral or negative, with the same rules. Answers that do not include your brand get no label.Sentiment score
Quotable summarizes sentiment as a score from 0 to 100:
On the Prompts table, Net sentiment above 70 reads as positive and below 40 as negative.
Example. Twenty passages about your brand: 12 positive, 6 neutral, 2 negative. Score = (1,200 + 300) / 20 = 75.
Comparing brands fairly
Competitors are analyzed from the same answers, with the same rules, so their scores are directly comparable to yours. Head-to-head on the Sentiment page puts your score next to one competitor’s as Favorability. A brand that few answers discuss has few passages, so read a gap between two small counts as a hint, not a verdict. For claims checked against facts you approve, see FactCheck.Next steps
Sentiment page
Brand perception, Head-to-head and strengths and weaknesses.
Visibility by topic
Which topics each competitor owns.