Answer Share

Answer Share is the share of AI answers on your topics that name or cite you, with its range. It is measured the same way for you and for every competitor, so the leaderboard puts everyone on one scale. Everything else on screen is either a reason behind that number or a lever on it.

What counts

An answer counts for a brand when it names the brand or cites one of the brand’s own pages. A brand that is named and cited in the same answer counts once.

Three things never count:

• A question that names the brand. “Fabrikam alternatives” or “Northwind vs Contoso” names those brands in the question, so the answer names them because the question did. That is not visibility, so the answer does not count for the brands its own question named. It still counts for every other brand in it. A brand that is only ever asked about by name on a topic has no share there, shown as a dash, not as zero.
• Answers without search. The poll’s asks with search off feed the Memory lens. Answer Share reads search answers only.
• No answer. An engine that failed, or a results page with no AI answer shown, counts neither for nor against anyone.

Weighting

The figure is built so that how often something was asked never tilts it:

• Questions weigh the same within a topic. The questions stand for the topic’s phrasings, so a question asked more often does not count for more of the topic.
• Repeat asks are discounted. Two asks of one question mostly agree with each other, so they count as less than two independent answers. The range reflects that.
• Engines weigh the same within a topic. An engine that answered more often does not tilt the topic’s share, and each engine’s share keeps its own range.
• Topics weigh by demand. Across topics, a high-demand topic counts more than a low-demand one. Until most of the site’s topics have a demand estimate graded B or better, the weights come from the demand tiers; after that, from the estimates themselves. The screen says which weighting is in use. See Topics and AI demand.

The range

Every share is shown with its range, for example “18% · 15–21”: the values the answers support. The range comes from the answers behind the share: how many there are, how many different questions they come from, and how much repeat asks of one question agree. Few answers, or few distinct questions, give a wide range. Ranges narrow as the poll adds answers.

A change is called only when it is clear. The Overview compares the last 28 days with the 28 days before; a change whose range includes zero reads “no clear change”. The Moves list holds clear changes only.

Leader and gap

• Leader: you or the rival with the highest Answer Share on a topic. A rival is a competitor, a brand you pinned, or another company with at least 5% of your Answer Share; an answer engine, a platform or a word for the category never leads. A topic where neither you nor a rival has a share has no leader.
• Gap: the leader’s Answer Share minus yours, in points.

Topics are sorted by demand and gap, because a wide gap on a high-demand topic is where a play is worth most.

Where in the answer

Answer Share counts a brand named sixth as it counts one named first. Four more figures say where a brand comes, read from the same answers (the order an answer names its brands and lists its sources):

• Named first: the share of the answers whose first named brand is this one.
• Average place: where the brand comes among the brands an answer names, on average (1 is first), over the answers that name it.
• First source: the share of the answers whose first cited source is one of the brand’s own pages.
• Source place: where the brand’s first page comes among an answer’s cited sources, on average, over the answers that cite it.

Named first and first source are read exactly like Answer Share (the same weights, range and change, and a question that names a brand does not count for it), so each is at most the brand’s Answer Share. The averages say how many answers they rest on.

How the answers speak of a brand

Being named is not being picked. Two more figures say how:

• Recommended in: the share of the answers that recommend the brand or name it their top pick, read like Answer Share, so it is at most the brand’s Answer Share. A brand named often but recommended rarely is one engines know and pass over.
• Tone: each sentence that names the brand is labelled from −1 (negative) to +1 (positive), and each question’s sentences are averaged. It reads neutral when it stays near zero; positive or negative when its whole range is on that side of zero; mixed when the range crosses zero. Tone is read for you and the competitors you track by domain, and shows from 5 labelled sentences. On Mentions, an answer that is clearly positive or negative about you says so.

Who counts as a competitor

Competitors are discovered from the answers: every brand an answer names or cites is recorded, with the domains that are its own. You can pin a brand, hide one, or add a competitor by name and domain. Hidden brands, the answer engines themselves and the platforms answers cite (review sites, forums, encyclopaedias) are left out of the leaderboard unless you show them. Your own brand is always on it.

Where it shows

• Overview: your share, its range and change, the leaderboard, the topics, the engines and the moves.
• Topics and each topic’s page: share, leader and gap per topic, per engine; each brand’s named first, average place and first source on the topic.
• Competitors: each competitor’s share and where it leads you, with its named first, average place, first source and source place beside yours.
• Results: your share since you started, against the leader, and how often you are named first (its line covers the last 90 days). See Results.

An unmeasured share is a dash, never 0. The design of the figure is also on the methodology page.