LLM Citation Quality measures not just whether a brand is cited, but how — in what context, with what sentiment, in what position in the answer. A brand cited negatively is retrieving but damaging. A brand cited in a hostile context is being amplified for the wrong reasons. Quality matters as much as count.
Dimensions: (1) Citation Context — is the brand cited favorably, neutrally, or adversarially? (2) Citation Position — is it the lead source or buried in a list of alternates? (3) Citation Sentiment — is the associated language positive, negative, or neutral? (4) Citation Accuracy — is the description accurate or muddled? (5) Citation Authority — is the source journal/publication tier 1 or tier 3?
Example: a brand with 100 total citations split as (80 favorable + 20 hostile) has stronger LLM Citation Quality than a brand with 100 citations split evenly (50 favorable + 50 hostile), even though total citation count is identical. The first brand's quality is 0.80; the second is 0.50.
In GEO and AI Communications measurement, Citation Quality is the emerging layer beyond raw Citation Share. A program optimizing for volume alone will generate citations that retrieve but damage. A program optimizing for quality alongside volume builds the kind of citations that actually drive perception shift and buyer preference. Quality tracking requires human audit and sentiment tagging, adding cost but critical for brand safety in the AI era.
