Imagine someone asks an AI assistant for a service like yours. Your brand appears in the response. That sounds promising—but is it a recommendation, a passing reference, or a warning? Did the answer use your website as a source? These are different questions, and each deserves its own answer.
A mention puts you in the conversation.
A mention means the answer refers to your brand. It may describe your offering, compare it with another business, or discuss a concern. Reading the surrounding sentence matters: presence alone does not tell you whether the answer helps a buyer consider you.
Illustrative example: “Aster Bikes sells commuter bicycles.”
Here, the fictional brand is named and described. The sentence does not recommend it. A statement such as “I could not verify Aster Bikes’ warranty” would also name the brand, with a very different implication.
A citation points to a source.
A citation is a link or source reference attached to the answer. It lets you inspect the material the response points to: your product page, a third-party review, a comparison article, or something else.
Your website can be cited without your brand being recommended. Equally, an answer can recommend a brand while citing another publisher. Keep those two views together: what the answer says, and which pages it uses to support the response.
A recommendation gives the brand a role.
A recommendation presents a brand as an option for the person asking. It may be qualified by budget, location, features or a particular need. Those qualifications are useful context, rather than detail to discard.
Illustrative example: “For a short city commute, consider Aster Bikes’ folding range.”
That is stronger than a descriptive mention. It connects the brand with a use case. It still does not establish that every buyer will see the same answer, or that the answer will lead to a sale.
Ask what sits behind the percentage.
A visibility percentage makes sense only alongside the questions that produced it. A brand might appear frequently when named in a question and rarely when someone asks an open question about the category. Both readings can be useful; they answer different business questions.
Before acting on a headline, look at the market, the question set, the platform, the measurement date and the answers themselves. A change in any of those can change what you are looking at.
Build a picture your team can use.
Read mentions, citations and recommendations together. Then ask: where are we being considered, which competitors appear, which sources shape the answer, and what do we want to understand next?
The useful outcome is a shared starting point for your team. A number helps you notice a pattern. The recorded answers help you decide what that pattern means.