Semantic branding: why positioning now needs to work for AI too.

Updated: March 18, 2026

In 1981, Al Ries and Jack Trout published a book that changed the way a generation of marketing professionals operated. “Positioning: The Battle for Your Mind” argued that the real competition wasn’t on the shelf or in the media plan, but in the consumer’s mind. To win, a brand needed to own a category. Volvo owned “car safety.” FedEx owned “next-day delivery.” Coca-Cola owned “soft drinks.” The brand that claimed the territory first had an almost insurmountable advantage.

For four decades, this model served us well. And it still does. But something fundamental has changed.

Today, when a consumer wants to make a purchase decision, they increasingly turn to AI . Not just search engines with AI-generated summaries from Google, but conversational tools like ChatGPT, Gemini, Claude, and Perplexity. They type questions like “what’s the best project management tool for a small creative agency?” and expect a direct answer. Not ten blue clicks, but a recommendation.

The question marketers need to ask now is uncomfortable, but urgent: what does their brand mean to the machines that are increasingly shaping consumer choices?

This is the challenge I call semantic branding , the discipline of shaping how major language models perceive, categorize, and recommend your brand .

The end of category ownership

Ries and Trout’s insight was that consumers have limited mental availability. They can only remember a handful of brands per category, so owning the category meant winning by default.

But LLMs don’t have limited mental capacity. They have virtually infinite memory and can process thousands of brand attributes simultaneously. More importantly, they don’t think in categories the way humans do. When someone asks an LLM for “a sustainable running shoe with good cushioning for under R$800,” the model isn’t searching in a mental category called “running shoes.” They’re constructing a bespoke category on the spot, defined by that specific combination of attributes.

This changes everything. In the age of AI , each consumer effectively creates their own category at the moment of consultation . The brands that are recommended are those that have built strong semantic associations with the right attributes, no longer those that own a single, static category position. This is what makes semantic branding fundamentally different from traditional positioning: it’s not about claiming territory, but about being semantically linked to the specific combination of attributes that each customer cares about.

From one position to multiple attributes

The practical implication is that brands need to migrate from owning a category to owning a constellation of attributes. Based on our experience, I recommend that brands identify approximately eight core attributes aligned with how their target customers describe their needs. They should be specific enough to differentiate, but broad enough to capture a significant volume of queries.

How do you know which attributes an LLM associates with your brand? You can start simply: ask. Ask ChatGPT, Claude, or Gemini to rate your brand against specific attributes using a scale of 0 to 100. In our tests, scores below 60 indicate weak association – the model barely connects your brand to that concept. Scores of 80 or higher suggest strong semantic ownership.

Try this exercise for your brand and your competitors. The gaps reveal your semantic branding opportunities and the attributes you should actively work to acquire.

The pages that really matter

This is where semantic branding differs from traditional performance marketing practices. For years, brand teams have focused their digital efforts on blogs, social media content, and top-of-funnel campaigns. Meanwhile, product pages, “About” sections, and institutional content were treated as secondary. They were merely hygiene factors to be maintained, not strategic assets to be optimized.

But when LLMs retrieve information to form recommendations, they give significant weight to the brand’s own institutional pages. If your product pages look like generic fact sheets, you’re leaving attribute ownership on the table. If your “About” page is corporate text full of fluff, you’re missing the opportunity to train how AI models understand what makes your brand unique.

Semantic branding is about treating each page as an opportunity to reinforce your strategically chosen attributes in a clear, consistent, and compelling way. The brands that will win are those that communicate their attributes not just in campaigns, but in every piece of content they control.

Branding is not dead and is more important than ever.

Every time a new technology promises consumers easy access to “perfect” information, someone predicts the death of branding. It happened with the performance revolution. It’s happening now with AI.

But the brand didn’t die back then, and it never will. In fact, the main factors for mentions in AI are linked to the brand: search volume for the brand, mentions of the brand on the web, and hyperlinks with anchor text (clickable text) for the brand.

What’s changing is where brand perception is formed . For decades, we’ve focused on building mental availability in human minds. Semantic branding adds a new dimension: model availability.

The fundamental questions of brand strategy—that is, what we want to be known for, who our audiences are, and what we want them to think and feel—remain exactly the same. What is new is that we have an additional audience to whom we must answer them: the language models that are increasingly mediating human choices.

Positioning has always been about winning a battle within the consumer’s mind. But today, before your brand reaches the mind, it may first go through an LLM (Life Model Trust) that decides whether or not to recommend it. Semantic branding is how you win both battles: the model battle and the mind battle.