Methodology

What it measures: Of all the companies an AI answer names, the weighted share that is your brand.
How it is calculated: Each company named in the answer counts once. The first company named counts one and a half times. Your share is your weight divided by the total weight. An answer that names no company scores zero.
What moves it: Being named in more answers, and being the first company an answer names.
What it does not mean: It is no ranking. When a rival is named first, your share stays below 100% even if you are the only other name.

Position-adjusted word count PAC

What it measures: How much of an answer talks about your brand, counting early sentences more than late ones.
How it is calculated: Each sentence that names your brand adds its word count, discounted by how far down the answer it sits. The total is divided by the answer's word count.
What moves it: More of the answer about you, and that text appearing near the top.
What it does not mean: It says nothing about tone. A long negative paragraph scores the same as a long positive one.

What it measures: The share of an answer's words that sit in sentences naming your brand.
How it is calculated: Words in sentences that name your brand, divided by all words in the answer. Position does not matter.
What moves it: More of the answer about you, wherever it appears.
What it does not mean: A high score with a much lower position-adjusted word count means you are discussed, but near the end.

What it measures: How much of your own vocabulary (product names, key terms) the answer uses.
How it is calculated: Your profile's key terms form a set. The score is the share of that set that appears anywhere in the answer. Saying a term ten times counts the same as once.
What moves it: Pages that use your product names and key terms clearly, so AI picks them up.
What it does not mean: It does not measure how often you are named, only which of your terms the answer used.

What it measures: Of the source links an AI answer cites, the share that point to your own domain.
How it is calculated: Links to your domain or its subdomains, divided by all cited links. A link whose page is gone (a 404 or 410, for example) is dropped from both counts. A link that blocks our check still counts.
What moves it: Pages AI can quote and chooses to cite for the prompts you track.
What it does not mean: Being named without a link does not count here. That shows up in share of voice.

What it measures: The average tone of the sentences that name your brand.
How it is calculated: Each sentence that names your brand gets a tone score from a fixed word list, so the same sentence always gets the same score. The scores are averaged. Error and refusal replies score nothing.
What moves it: What answers say about you: reviews, comparisons and the sources AI reads.
What it does not mean: It does not judge whether the answer is correct.

What it measures: One number that blends being seen with being liked.
How it is calculated: Half AI share of voice plus half sentiment.
What moves it: Either input. Read the two inputs to see which one moved.
What it does not mean: A well-known brand with poor tone and a little-known brand with good tone can land on the same number.

What it measures: Whether your brand is among the first three companies an answer names.
How it is calculated: Your mentions among the first three companies named, divided by how many companies that is.
What moves it: Being named earlier in answers.
What it does not mean: It says nothing about the rest of the answer.

What it measures: Whether AI brings up your brand on its own, when the question does not name you.
How it is calculated: Only category questions with your brand name removed are asked. The score is your mentions divided by your mentions plus your competitors' mentions.
What moves it: Being known for your category across the sources AI reads.
What it does not mean: A question that already names your brand never counts here.

What it measures: How easy your pages are for AI to quote, from 0 to 100.
How it is calculated: A weighted blend of three page factors: citability (42%), structure for AI retrieval (33%) and multimodal content (25%).
What moves it: Clear answer-first content, clean page structure and multimodal content.
What it does not mean: It reads your page HTML only. It does not say whether AI cites you today.

What it measures: SEO, schema, security, mobile, URL, Core Web Vitals, robots.txt and sitemap, each from 0 to 100.
How it is calculated: Each score starts at 100 and loses points for every problem found, more for a critical problem than a minor one. It never goes below zero.
What moves it: Fixing the problems listed in the technical fix list.
What it does not mean: The same HTML always gets the same score. A change means the page changed or the method version did.

What it measures: How a page shows experience, expertise, authority and trust, each from 0 to 10.
How it is calculated: An AI model grades each of the four on a fixed 0 to 10 scale, with a written reason. The overall score is the average of the four, computed in code.
What moves it: Named authors, first-hand detail, cited sources and clear contact and company facts.
What it does not mean: An AI model makes this judgment, so the same page can score a little differently on another run.

What it measures: When ChatGPT answers with a shopping carousel, whether your products appear in it.
How it is calculated: A count of answers: of the answers with a carousel in your market, how many show your brand.
What moves it: Product feeds and listings that name your brand and products clearly.
What it does not mean: It is a count, never a score, and it feeds no other number.