By Joao da Silva and Maryanna Franco (BrilliantSEO) ยท June 6, 2026
TL;DR. A category isn't a flat list to an AI. It's a cluster of brands that get mentioned together, with one brand at the dead center. We call that center the Category Prototype. For athleisure it's lululemon: named first in most answers (average position 1.64) and co-mentioned with Alo Yoga 534 times, Nike 482, Gymshark 264. New Balance co-occurs with lululemon 21 times, an unconnected node outside the cluster. To get recommended, you don't have to beat the prototype. You have to enter its cluster.
When an AI answers "what are the best athleisure brands?", it isn't reading down a ranked list. It's pulling from a dense web of brands that tend to appear together in the content it learned from. That web has a shape, and at the middle of it sits one brand the model treats as the definition of the category. Everything else is positioned by how close it sits to that center.
This is the third mechanism in the Beyond KG Strength series. Category Coding decides which cluster you belong to. This piece is about your position inside the cluster once you're in it, and why the brand at the center has such a gravitational pull.
What a Category Prototype is
A Category Prototype is the single brand an AI treats as most representative of a category. It's the one that comes to mind first, gets named first, and anchors the list everything else hangs off. In cognitive terms it's the same idea as "robin" being the prototype for "bird": when you think of the category, you think of that member.
For athleisure, lululemon is unmistakably the prototype, and the data shows it two different ways: it's named more than any other brand, and it's named first.
The cluster has one brand at its center
The clearest signal is co-occurrence: when the models recommend brands together, who appears alongside whom. lululemon sits at the center of a tight cluster. The brands that share the athleisure box co-occur with it constantly. The footwear-coded brands don't.
Alo Yoga is named with lululemon 534 times. New Balance, the strongest brand in our whole sample by Knowledge Graph score, co-occurs with it 21 times. That's the difference between being in the cluster and being a node floating outside it. New Balance isn't competing with lululemon for athleisure recommendations. As far as the models are concerned, it isn't in the conversation at all.
The prototype gets named first
Co-occurrence shows who's in the cluster. Position shows who's at the center. We measured the average ordinal position at which each brand appears in recommendation answers, where 1 means it was the first brand named. lululemon's average is 1.64: in most responses where it appears, it's named first or second. Nothing else is close.
Being named first is not a vanity metric. The order brands appear in an AI answer is the order a buyer reads them, and the first name carries disproportionate weight. The prototype gets that slot by default, every time the category comes up.
The category AI sees is bigger than your competitive set
Here's the part that reframes how you think about competitors. The models don't restrict themselves to the obvious names. Our extractor captured more than 260 distinct brands across recommendation answers, and two we never put in the study outranked most of the ones we did: Athleta (74% of recommendation runs) and Vuori (71%) were each mentioned more often than 9 of our 12 target brands (study dashboard).
These two sit between Alo Yoga and Nike on both mention rate and position. They are co-anchors of the cluster, and most brand teams wouldn't list them first if you asked who lululemon's competitors are. The lesson: the category AI recommends from is wider and more specific than the competitive set in your deck. Before you try to enter a cluster, you have to know who AI thinks is actually in it.
How to enter the cluster
You don't dislodge the prototype. lululemon's position is the product of years of accumulated coverage, and chasing it head-on is a losing game. What you can do is become a recognized member of the cluster, the way Athleta and Vuori are. Three steps:
- Map the real cluster. Run several "best [category]" prompts across ChatGPT, Gemini, Claude, and Perplexity. Write down every brand named and the order. That list, not your competitive deck, is the cluster AI sees. Note who's named first (the prototype) and whether you appear at all.
- Measure your distance from the center. Search the third-party web for content that names the prototype, listicles, comparisons, "brands like [prototype]" round-ups. Are you in those pieces? Co-mention with the prototype in independent content is what puts you in the cluster.
- Earn co-mention, not self-mention. The way in is getting placed alongside the category leader in content you don't own: category round-ups, comparison articles, lifestyle and trade publications. This is the same mechanism behind The Coverage Gap, recommendation runs on third-party coverage, and the bridging move in Sub-Stream Strength. (slugs provisional)
The prototype defines the category. Your job is to get close enough to it, in the eyes of the content AI reads, that you're part of the same answer.
Start with the pillar for the full picture: Brand Strength Gets You Recognized, Not Recommended.
FAQ
What is a Category Prototype? It's the single brand an AI treats as most representative of a category, the one named first and most often when the category comes up. For athleisure it's lululemon, with an average mention position of 1.64.
Why does the first brand named matter? AI answers are read top to bottom, so the first brand carries the most weight with the buyer. The prototype occupies that slot by default whenever the category is queried, which compounds its advantage over time.
How do I get my brand into the category cluster? Earn third-party content that co-mentions you alongside the prototype: category round-ups, comparisons, and "brands like [leader]" articles. Co-mention in content you don't own, not your own website, is what AI reads as cluster membership.
Who decides which brands are in my category? The published web does, and it may not match your competitor list. In our athleisure data, Athleta and Vuori outranked most pre-registered brands despite not being in our set. Map the cluster with live prompts before assuming you know it.
Part of the Beyond KG Strength series (Franco & da Silva, 2026, DOI: 10.5281/zenodo.20331344). Pillar: Brand Strength Gets You Recognized, Not Recommended. Previous: The KG-Hijacked Entity.