The Quellan Answer

The Gap After the Diagnosis

A Quellan diagnosis tells you how AI systems describe your brand. The Quellan Method explains how we measure it. This document explains what we have learned from watching.

Every diagnosis we publish produces two reactions in the room. The first is recognition. CMOs see the score and the framing and immediately know it is true. The second, usually one breath later, is the question. What do we do about it.

For most of the last two years, the honest answer was: "Talk to us, we will figure it out together." That is consulting. It is not a product, and it is not a position. If you cannot say what you believe without a statement of work, you do not have a worldview. You have a services invoice.

This is the worldview. It sits next to The Method because the pair is the product. The Method is how we see. The Answer is what we see.

Most brands think AI visibility is a distribution problem. It is a representation problem. The fix looks different when you name it correctly.

We have observed five patterns in almost every diagnosis. They are not independent. They compound. A brand weak in one dimension is usually weak in two or three. Naming them is not a checklist. It is a grammar. Once you have it, you can diagnose without us.

The Corroboration Problem

The Framing Problem

The Specificity Problem

The Culture-to-Category Problem

The Source-Shape Problem

The Corroboration Problem

AI systems do not trust what your brand says about itself. They trust what multiple independent parties say about you, repeatedly, in editorial voice.

Brands operate on a media model: spend the right money, earn the impressions, grow the numbers. AI operates on a corroboration model: find the facts two or three independent sources agree on, surface those facts, ignore the rest. Two completely different retrieval logics, and most marketing budgets are pointed at the wrong one.

When Oatly appears in AI answers about oat milk, it appears behind Califia Farms and Chobani. The reason is not taste. Oatly sits in the corpus constantly. It loses the recommendation because Serious Eats, Food & Wine, and Simple Green Smoothies ranked other brands first in independent taste tests. AI does not know which brand is louder. It knows which brand is named more often by parties with no skin in the game.

Nike has the inverse problem. Its owned content, press, and campaign coverage flood the corpus. But when the query is "best running shoes," specialty-running editorial corroborates ASICS, Brooks, Hoka. Nike's category presence is overwhelmed by its category framing. Present constantly. Recommended rarely.

The principle

You become visible by being talked about, not by talking about yourself.

The test for whether a mention counts is simple. Would this sentence exist if you did not pay for it, write it, or brief it. If the answer is no, the sentence does not enter the corroboration layer AI relies on. You can have a thousand such sentences and still not move a single AI recommendation.

Observed in Oatly (Analysis 001) Nike (Analysis 002)

The Framing Problem

AI does not report that you exist. AI reports in what verb tense you exist.

When an AI system is asked about your brand, it does not return a description. It returns a position. The position is built from how the corpus talks about you: whether you are described as currently doing something or historically having done it. That difference is the difference between being the recommendation and being a reference note.

Nike, in Claude and ChatGPT responses about athletic footwear, is consistently described in past-tense strengths. Heritage. Scale. The largest sportswear brand. The company behind Air Jordan. These are institutional facts. Competitors, meanwhile, are described in present-tense momentum. On Running is "doing interesting work." Hoka is "rapidly growing." New Balance is "experiencing a renaissance." Same category. Completely different verb tense. Completely different AI outcomes.

Oatly shows a second kind of framing problem. When it appears, it is framed by its brand archetypes: jester, outlaw, sage. These are culturally alive. They are also algorithmically inert. A model trained to retrieve product recommendations cannot weigh wit. It can only read whether someone said this product is the best at something specific. Oatly's framing compresses into "known but not recommended."

The principle

AI reads whether you are a subject or an object of the current sentence.

Brands are described in two modes. Past institutional (was founded, built, dominated, pioneered) or present participatory (is doing, is trying, is reshaping, is winning). AI rewards the second and files the first as context. The work is not to claim the present tense. The work is to get cited in the present tense by the sources AI reads.

Observed in Oatly (Analysis 001) Nike (Analysis 002)

The Specificity Problem

AI cannot read vibes. It reads claims. Vague equity is invisible. Specific claims survive compression into an AI answer.

Most brands describe themselves in adjectives: premium, innovative, iconic, authentic, category-defining. These words do not exist in AI answers. Not because the AI is filtering them out. Because they do not compress. A recommendation engine cannot pass "premium" to the next layer of reasoning. It has to turn every input into something structured enough to be retrieved, compared, and weighted. Adjectives do not retrieve.

This is why Oatly's Barista Edition survives in AI and Oatly's brand narrative does not. "Best oat milk for lattes" is a specific claim. Multiple independent sources agree. The claim retrieves. "Post Milk Generation" is a cultural movement. It does not retrieve, because AI does not know which facts to attach to it. Same brand. Two different fates.

Nike's "return to sport" strategy shows the same pattern in reverse. The strategy itself is vague (every sportswear brand returns to sport). The artefacts are specific. The Alphafly is a specific running shoe with specific technology and specific test data. It survives AI answers. The broader cultural thesis of Nike's correction does not, because AI has no place to put it.

The principle

AI retrieves what can be verified, compared, and cited. Nothing else exists.

Every brand claim should be tested against one question: if a stranger wrote this sentence about a competitor, would you recognise it as factually accurate. If the answer is "it could apply to any brand in the category," the claim is invisible to AI. If the answer is "no, that would be wrong," the claim is a candidate for retrieval. Specificity is not a style choice. It is the price of entry.

Observed in Oatly (Analysis 001) Nike (Analysis 002)

The Culture-to-Category Problem

Cultural equity lives in culture. AI retrieves in categories. The translation has to happen deliberately, and most brands never notice it must.

A brand builds cultural resonance through consistent voice, recognisable design, and repeated cultural contribution. Humans translate that resonance into purchase intent through recognition: I know this brand, I trust this brand, I want this brand. The mechanism is cultural.

AI does not make that translation. It cannot. AI is asked a category question ("best oat milk," "best running shoes," "best luxury bag") and must retrieve a category answer. Cultural equity does not map to category facts unless someone has done the mapping work: written the editorial, run the comparison, named the cultural brand specifically as the best for this specific use case.

Oatly lives this problem publicly. The brand is a cultural phenomenon. Every CMO in the world has referenced Oatly in a strategy deck. That equity exists in the corpus as cultural commentary. It does not exist as category recommendation. When someone asks an AI for the best oat milk, the AI does not read a decade of cultural salience. It reads category-specific editorial and returns Califia.

Nike shows the same gap with higher stakes. Nike the cultural phenomenon is unmatched in sportswear. Nike the category answer, when the category is running shoes and the judge is editorial consensus, is not the first name mentioned. The cultural equity is intact. The category translation has not been done.

The principle

Brands are built in culture. They are retrieved in categories. The bridge between the two is editorial.

The translation work is not a campaign. It is a sustained investment in the kind of editorial coverage that takes a cultural brand and positions it specifically against a category question. "Oatly is a design icon" is cultural. "Oatly Barista Edition is the best oat milk for lattes" is category. Both need to exist in the corpus, or the cultural equity stays where it was built and never reaches where it is retrieved.

Observed in Oatly (Analysis 001) Nike (Analysis 002)

The Source-Shape Problem

AI does not cite brands. It cites sources with a specific shape: independent, structured, repeated, third-party-written. You become visible not by producing more content, but by becoming the kind of entity such sources write about.

This is the pattern that makes the other four actionable. If the first four name what is broken, Source-Shape names the structural repair. It is the hardest pattern to see, because it is never about a single piece of content. It is about the texture of the coverage your brand is the subject of.

AI retrieval systems learn, over millions of training examples, what a trustworthy source looks like. A Wikipedia article. A Financial Times feature. A Serious Eats taste test. A Gartner quadrant. A peer-reviewed paper. A Reddit thread with multiple commenters agreeing. The shape is consistent: the source is independent of the subject, the writing is structured around an evaluative claim, and similar shapes repeat across venues.

Brands that become visible in AI answers have one thing in common. They are the kind of entity that gets written about in that shape, by people who do not work for them. Oatly Barista Edition exists in AI answers because Serious Eats and the coffee press corroborate it in reviews. Nike's Alphafly exists because running-specialist editorial treats it as a product worth comparing. The cultural brand wraps these artefacts, but it is the artefacts that retrieve.

This reframes the work. The game is not to produce more brand content. The game is to produce the kind of specific, testable, claim-bearing product, gesture, or idea that source-shaped content gets written about.

The principle

Stop producing content. Start producing artefacts that content gets written about.

The question a brand should ask itself, every quarter, is not "what should we say next." It is "what should we do next that an independent source will have a reason to describe." The answer is rarely a campaign. It is usually a product decision, a product detail, a specific cultural gesture, a measurable claim. AI is a retrieval system. It rewards brands that generate things worth retrieving about, and ignores brands that only generate things to say.

Observed in Oatly (Analysis 001) Nike (Analysis 002)

What the Patterns Add Up To

The five patterns are not a checklist. They are a grammar. Once you can speak it, you can diagnose your own brand without us.

Each pattern describes the same underlying condition from a different angle. AI systems build representation from signals they trust: corroborated, present-tense, specific, category-anchored, source-shaped. A brand is visible to AI when its presence in the corpus matches that shape. A brand is invisible when its presence does not, regardless of how strong the brand is in the culture.

This is why the AEO and GEO vocabulary currently circulating in the SEO industry misses the point. Those frames carry ranking logic from search into retrieval. They assume the problem is getting crawled more often or being worded more tightly. The problem is not distribution. The problem is representation. The brand is being represented in AI answers right now. The question is what that representation actually says. The patterns are how we name what it says.

The fix is not technical. It is not an SEO intervention or a content production push. The fix is a decade of editorial patience combined with the right kind of brand gestures: specific, verifiable, worth writing about. Most brands are not set up to do that work. The ones that are will compound. Everyone else will slowly become invisible, without ever knowing when the visibility left.

SEO measured ranking. The Method measures representation. The Answer describes what representation actually is.

This document will change. The patterns here are drawn from the first analyses we published. As the sample grows, we expect to revise, rename, or split them. We will date every revision. We will not pretend the early version was the final version.

What will not change is the frame. AI is not a search engine with a new interface. It is a representation engine, and what it represents about your brand is decided by signals you mostly do not control yet. The Answer is our best current articulation of what those signals are and how they work. Use it. Argue with it. Cite it. If you spot a pattern we missed, we want to hear about it.

The Quellan Answer.

The Gap After the Diagnosis

The Corroboration Problem

The Framing Problem

The Specificity Problem

The Culture-to-Category Problem

The Source-Shape Problem

What the Patterns Add Up To

Method measures. Answer interprets. Analyses prove.