How AI assistants decide which brands to mention
Mention behavior is provider-dependent and signal-dependent. The methodology distinguishes mention from citation as separate axes, classifies sentiment per brand per response, and gives weight to list position. The findings below summarize what the test pipeline measures consistently.
Citation and mention are independent signals
Some providers cite source domains without naming the brand in prose; others name the brand without citing. The engine measures both axes. Citation reach for the target brand is computed independently of name-mention frequency, and competitor URL citations are recorded as context without contributing to the citation score.
Sentiment is classified per brand per response
When a single response names multiple brands, each receives its own sentiment label (recommendation, neutral, negative-qualified, negative-blanket). A second model handles classification so the graded provider does not also grade itself. This preserves the sub-signal that drives competitive positioning.
List position carries disproportionate weight
Position 1 in an ordered list appears in a substantially higher share of useful answers than positions 2 through 4. The rubric applies diminishing weights from position 1 outward. First-mention dominance is the most consequential structural lever for visibility scores.
Citability signals are observable and improvable
Providers favor sources with structured data (JSON-LD), unambiguous robots directives, stable URL patterns, and published methodology references. These are configurable site-side. Domain authority compounds the effect, but baseline crawlability and structured markup are the threshold conditions.
Run the analysis to see your brand's score across the four providers.