What Is Mention Rate?
Mention rate is the percentage of relevant AI prompts for which your brand is named anywhere in the generated answer. Run a category prompt set across engines; count the runs where the answer says your name; divide by total runs. It is the broadest AI visibility metric — the "were we in the room at all?" number — and the natural starting KPI for any GEO program.
How do mention rate and citation rate diverge?
They measure different assets and routinely tell opposite stories:
| Scenario | Mention rate | Citation rate | Diagnosis |
|---|---|---|---|
| Famous brand, weak content | High | Low | Model memory carries you; third parties source the claims |
| Unknown brand, strong content | Low | High | Your pages get cited, but the brand isn't in consideration sets yet |
| Healthy GEO position | High | High | You're recommended and you control the sources |
The gap between the two is itself diagnostic. High mentions with low citations means engines describe you using other people's words — review sites, old articles, competitor comparisons. Low mentions with high citations means your content earns trust but your entity hasn't; the fix is brand-building, not more pages. Segmenting both per prompt, as visibility tracking platforms do, turns the gap into a roadmap.
Why must mention rate be sampled, not spot-checked?
Because generated answers are non-deterministic. LLM decoding uses temperature-based sampling, so the same unbranded prompt can name four brands in one run and a different four the next. A brand's true mention rate is a probability, estimated by repeated runs — practitioners average across multiple runs per prompt per engine per week before trusting the number. Spot-checking one answer and concluding "we're in ChatGPT" is the most common measurement error in the field, and the first habit proper tracking replaces.
What lifts mention rate?
Mentions are driven more by consensus than by your own site: engines name brands that appear repeatedly across retrieved sources — best-of lists, review platforms, community threads — and that models recognize from training. The levers: earn placement in the ranked lists engines cite, build review-corpus depth on G2/Capterra-class sites, sustain community presence, and keep entity descriptions consistent everywhere so mentions consolidate rather than fragment.
Example
A video-hosting startup measures 6% mention rate on 120 unbranded prompts. Analysis shows engines cite three "best video platform" listicles it is absent from. Placement in two of them lifts mention rate to 21% in eight weeks — no changes to its own site required.
Frequently asked questions
- What is the difference between mention rate and citation rate?
- Mention rate counts answers where your brand is named in the text; citation rate counts answers where your domain appears as a source. A brand can be mentioned from model memory without its site being cited, and a domain can be cited in an answer that never names the brand.
- Should mention rate be measured on branded or unbranded prompts?
- Unbranded. Being mentioned when the user already typed your name proves little. Mention rate on unbranded category prompts — 'best X', 'how do I solve Y' — measures whether AI engines put you in consideration sets you didn't buy your way into.
Keep exploring
See how AI engines talk about your brand — track mentions across ChatGPT, Perplexity, Claude, Gemini and 5 more. Start with Menra