AI NativeEvery answer a scan reads gets placed on a five-rung outcome scale. From the bottom up: absent, mentioned, listed_option, recommended, recommended_first. That scale is the visibility ladder, and it is the most precise way to read a scan result because it tells you not just whether you appeared, but how strongly you appeared.
What the five rungs mean
Absent means the answer contained no trace of your brand. No name, no domain, no recognisable alias. You were not part of the conversation at all.
Mentioned means your name appeared in the answer but the text did not treat you as a choice. A passing reference, a comparison footnote, a name alongside five others with no verdict: all of these land here. Mentioned is presence without preference.
Listed option means the answer placed you among a set of options the user could consider. You are in the list, acknowledged as a real contender, but the answer does not endorse any one option over another. This is a meaningful step up from being merely mentioned.
Recommended means the answer put you forward as a good choice. The AI said you are worth choosing, framed you as a credible option for the user's stated need. This is where preference begins.
Recommended first means you were the top recommendation. The answer led with you, named you as the strongest or most suitable choice, and placed every other option behind you. This is the position buyers remember.
Why a ladder, not a score
A single binary (present or absent) would lose most of the information. An answer that buries your brand in a paragraph of caveats is different from one that opens with your name. A ladder captures that difference without turning it into a spurious decimal. The rungs are an ordered qualitative scale: higher is better, and the distance between rungs is meaningful.
The ladder mean is the summary number you see per cell in the dashboard. It converts the five rungs to a 0 to 1 scale (absent = 0, recommended_first = 1) and averages across all answers in the cell. A ladder mean of 0.5 means you are sitting roughly at the listed_option to recommended boundary on average, which is a very different situation from a mean of 0.25 (trending toward mentioned) or 0.75 (trending toward recommended_first).
How classification works
Presence is determined by a deterministic check before any AI judgment is involved. If your brand name, a recognised alias, or one of your known domains appears anywhere in the answer text or its citations, the answer cannot be classified as absent. That floor is reliable: it does not depend on a language model's interpretation.
Once presence is confirmed, an AI judge reads the answer against the question asked and decides which rung above absent the answer sits on. The judge applies a rubric: it checks whether the answer endorsed you, how prominently, and whether that endorsement came with a position. Because this is a judgment call on natural language, the same answer type can occasionally land on different rungs at different runs. That is expected behaviour; running each prompt several times and averaging is what converts individual judgments into a stable rate.
Reading your ladder position
A high ladder mean in a cell tells you that when this persona asks this kind of question on this engine, you are coming out well. A low mean tells you the opposite. The gap between a cell where you average recommended and a cell where you average mentioned is the gap between two different buyer decisions.
The most actionable reading is always the comparison: your ladder mean versus your competitors' ladder mean in the same cell. Share of voice and the ladder mean together tell you whether you dominate the answers in a cell or share them with rivals who sit just as high on the ladder as you.
Questions
What is the difference between mentioned and listed_option?
Mentioned means your name appeared in the answer without the AI treating you as a choice the user should weigh. Listed option is a step up: the answer placed you among options the user could actively consider, even if it did not endorse one over another.
Can an answer be absent even when my website is cited?
No. A citation from your domain counts as evidence of brand presence. If an answer cites your site, it cannot be classified as absent regardless of whether your brand name appeared in the prose. The presence check covers text, citations, and any brand domain that appears in the answer.
Why does the ladder mean sometimes move between scans when I have not changed anything?
AI answers are stochastic. The same question asked twice can produce a different answer, and a different rung classification. Your ladder mean is an average across multiple runs, which smooths this out, but small movements between scans are expected noise. A sustained directional move is signal; a single-scan wobble is not.
How do I move from listed_option to recommended?
The gap between being listed and being recommended is almost always about the sources behind the answer. AI answers draw on content that positions you as a credible, specific solution for the question being asked. The scan shows you which domains are cited in answers where you are recommended versus merely listed. That is where the work is: getting those sources to carry the evidence that turns a listing into an endorsement.
Does the ladder apply to both branded and unbranded prompts?
Yes. Branded questions are about your brand directly, so the brand is almost always present and the ladder starts at mentioned at minimum. Unbranded questions ask about a category without naming you, so absent is a real possible outcome. Reading ladder position separately for branded and unbranded prompts tells you two different things: how well you are represented when buyers already know your name, versus how often you enter the conversation when they do not yet.
Where can I read about mention and recommendation rates?
See Mention versus recommendation for how AI Native converts the ladder into the two headline rates, and AI share of voice for how your ladder position compares to competitors across the same answers.