Definition
An AI mention is an instance where an AI answer engine names a brand, product, person, or competitor in the text of a generated answer, whether or not a source URL is attached. It is the answer referring to the entity in prose: “for small teams, consider Brand X” is a mention of Brand X even if no link accompanies it.
Where Mentions Come From
A mention can originate from two places, and diagnosing which one is important for fixing gaps. The model may name a brand from training data: accumulated public coverage, documentation, reviews, and press that taught it the brand belongs to a category. Or the mention may come from retrieval: the engine fetched pages at answer time, and those pages named the brand. Training-data mentions are durable but slow to change; retrieval-driven mentions can appear within weeks when crawlable pages, owned or third-party, clearly associate the brand with the question being asked. A brand that is rarely mentioned usually has an entity problem: the public web does not state plainly and consistently what it is, what it does, and which category it belongs to.
Mention vs Citation
A mention and an AI citation describe different events, and confusing them leads to the wrong fixes. A mention names the brand in the answer text; a citation attaches a source reference, usually a URL, to a claim. The four combinations each mean something: mentioned and cited is the strongest position; mentioned but never cited means the model knows the brand but does not treat its pages as evidence; cited but not named means an owned page supports a claim without the brand entering the conversation; neither means the brand is invisible for that prompt. Mentions reflect what engines believe about a brand; citations reflect which pages they trust. Both feed AI visibility reporting.
A Worked Example
A hypothetical example: an email-security vendor runs 30 buyer prompts across four engines, 120 answers per run. Its brand is named in 42 answers, a 35 percent mention rate. Segmenting shows the rate is 60 percent on prompts containing the vendor’s product name but only 12 percent on generic category prompts like “best phishing protection for law firms,” where two competitors dominate. The generic-prompt gap is the actionable finding: the engines do not yet associate the brand with the category, so the work is category-level content, consistent entity naming, and presence in the roundups those answers cite, not more branded content.
How To Measure AI Mentions
Track mention rate over a stable prompt set, repeated on a cadence, segmented by engine, market, and competitor set. Record who else is mentioned in the same answers and how the brand is framed, because a dismissive mention (“X exists but lacks enterprise features”) is a different situation from a recommendation. Never read a single response as proof; outputs vary with phrasing and retrieval state, so the rate and its trend are the signal. In AEO Goal, AI visibility tracking runs this measurement across ChatGPT, Claude, Gemini, and Perplexity, and the method is documented in the AI visibility tracking methodology.
How To Improve Mention Rate
Start from the entity problem: strengthen the canonical pages that state what the company is, what the product does, and which category it competes in, using the same names everywhere, reinforced with accurate structured data. Then work the retrieval door: publish direct answers to the category prompts where the brand is absent, and earn presence in the third-party sources the engines already cite for those prompts. Comparing where competitors are named for prompts you should own is the natural starting point, which is what competitor AI visibility surfaces.
Where It Fits In An AEO Program
Mention rate is usually the earliest-moving signal in an answer engine optimization program: engines often begin naming a brand before they cite its pages, so a rising mention rate with a flat citation rate is a normal intermediate state, not a failure. Aggregated against competitors, mention data becomes Share of Model, the executive roll-up of category presence, and paired with citation data it separates “the engines know us” from “the engines trust our pages.”
Related Terms
Related terms include AI citation, AI visibility, Share of Model, LLM visibility, and brand monitoring.