Mention Rate
- Mention Rate is presence, not authority or exclusivity.
- Count span-aware, or substring matches inflate it.
- Never report it alone — pair it with Citation Rate or Brand Recommendation Share.
Consensus definition
Across LLM-visibility practice, Mention Rate is a binary presence metric: for each answer in a prompt set, the brand is either named or not, and the rate is the average over the set. It is widely reported by AI brand-monitoring tools as a first-line visibility number1, as AI answers become a measurable surface in their own right2.
rhinegold operator caution
Rhinegold's caution is threefold. First, the count must be span-aware: a contiguous brand mention, not a substring that happens to appear inside an unrelated word, which would inflate the numerator. Second, a mention without a source link is presence, not authority — and Mention Rate overstates standing when a brand appears mostly in long aggregate lists, which is exactly why it should be read alongside Brand Recommendation Share. Third, the headline number is silently inflated whenever the prompt set mixes prompts that name the brand with prompts that do not — a mention in a brand-aided prompt is closer to a tautology than to a competitive signal. The split between unaided recall and aided recognition is canonical in brand-equity research and should be carried into GEO measurement: see Aided vs Unaided Brand Recall in GEO.
Operational use
Mention Rate is a broad, robust presence indicator — especially early, when few source citations exist yet. It is well suited to tracking visibility trends over time across a stable prompt set, and it is the headline volume signal carried by disciplined LLM Brand Tracking programmes.
Measurement boundary
Being binary, Mention Rate says nothing about exclusivity (that is Brand Recommendation Share) or about whether the brand was cited as a source (that is Citation Rate). It is volatile on small prompt sets and sensitive to brand-name ambiguity.
Distinct from
Against Citation Rate, which requires a source or URL backing the mention. Against Brand Recommendation Share, which weights the mention by the number of competitors named alongside it. Against Share of Voice, which expresses presence relative to the competitor set rather than in absolute terms. Against AI Mention Velocity, which captures the rate of change of mention count over time rather than its level. Against Multi-Axis Sentiment in GEO, which scores the tone and hedging of the statement carrying the mention rather than presence itself.
Operational note
Common mistakes
- Counting substring matches instead of span-aware mentions, which inflates the rate.
- Reading Mention Rate as evidence of authority — that is what Citation Rate is for.
- Averaging over the wrong prompt set, e.g. mixing in non-discovery prompts.
- Reporting a single aggregate Mention Rate over a prompt set that mixes brand-aided and unaided prompts — see Aided vs Unaided Brand Recall in GEO.
Where consensus is missing
While the concept is established, the exact counting rules — span-aware matching, alias and parent/subsidiary handling — are not standardized across tools, so cross-tool Mention Rate numbers are not directly comparable.
Sources & deeper reading
- rhinegold Insights, Episode 04 — How to Measure What You Can't See (GEO metrics for B2B)
- rhinegold span-aware brand-detection methodology and cross-provider mention benchmarks — shared with engaged clients and partners.