When ChatGPT or Claude or Perplexity cites your brand, the citation is rarely just "you exist". It is normally one of four shapes:
- Positive - explicit recommendation, "best", "leading", "trusted".
- Neutral - factual description or listing among many without preference.
- Negative - warning, complaint, "avoid", "outdated".
- Comparative - mentioned alongside competitors with explicit pros / cons framing.
v0.8 classifies the surrounding paragraph automatically (Pro and Agency tiers only). The result is cached on a sha256 of the paragraph, so the same passage cited twice never re-bills Claude. The chip on the citation row tells you which bucket each mention fell into.
Why it matters: a comparative-positive mention is gold (you are in the consideration set); a comparative-negative one is a content gap that the brief generator can address with a head-to-head section.