If you’ve spent any time near a marketing function in the last decade (especially in CPG/FMCG or Beauty), you have almost certainly had Byron Sharp’s How Brands Grow pressed into your hands at some point. It is one of those books that quietly rewired how serious marketers talk about their job. And the idea inside it that has done the most work — the one that genuinely changed planning conversations in boardrooms — is mental availability.
The argument, briefly. Most category purchases are low-attention. People are not weighing brands against each other in any deliberate way. They are walking down an aisle, scrolling a feed, scanning a shelf, and the brand that comes to mind quickly, easily, and in the right buying situation is the brand that gets bought. Mental availability is the probability that your brand is thought of in a buying moment. Distinctive assets — the colours, the logo, the sonic cue, the spokesperson — exist to make that recall easier. Reach exists to make sure the memory structure is built across enough of the category in the first place.
It is a beautifully practical theory. It explains why famous brands keep winning even when their products are not obviously better. It explains why penetration matters more than loyalty. It explains why distinctive assets are worth defending with something approaching religious fervour. And for twenty years, it has been roughly correct.
The question I want to pose today, is whether the conditions that made it correct are still the conditions we are operating in.
The quiet assumption underneath the theory
Mental availability is a theory about human memory in a buying moment. It assumes the human is the one doing the remembering. It assumes the buying moment is the moment of choice. It assumes the brain in question is, in the relevant sense, yours.
Increasingly, none of those things are reliably true. Or at least, they’re questionable.
When someone asks Gemini for the best running shoes for flat feet, or Claude to compare three accounting software options, or Perplexity to recommend a holiday destination for a family of four with a tight budget, something interesting is happening. The shopper has not stopped thinking. But they have outsourced the recall step. They are no longer scanning their own memory for the brands that come to mind. They are scanning a model’s response.
That is not a small change. The model has, in effect, perfect recall. It does not forget brands the way humans do. It does not have decaying memory structures. It does not preferentially remember the brand whose ad it saw most recently. It has read more about your category than any human shopper ever will, and it will surface what its training and retrieval suggest is most relevant to the question — which is not the same thing as what is most famous.
Which means the mechanism by which mental availability has historically converted into sales — the human’s brain reaches for the easy, familiar option in a low-attention moment — gets disrupted at exactly the point it was designed to fire.
So is mental availability dead?
No. But its centre of gravity is arguably moving, and the move matters.
Mental availability still works perfectly well in the moments where humans are still doing the remembering — which, to be clear, is still the vast majority of category purchases today, and will remain so for some time. Walking down a supermarket aisle. Scrolling a feed. Hearing a friend mention they need a plumber. The mechanism Sharp described is intact in those moments, and the brands that have invested in distinctive assets and broad reach are still going to win them.
What is changing is that a growing share of high-consideration category entry points — the ones where someone explicitly asks for a recommendation — are being intercepted by a system that is not playing the mental availability game. And in those moments, you are not competing for the shopper’s memory. You are competing for the model’s.
That is a different competition. It rewards different things. And it requires marketers, and the people responsible for building marketing capability, to hold two ideas at once.
Holding two ideas at once
The first idea is that mental availability, as classically understood, is more important in the moments where humans are still the decision-makers — not less. Because if a meaningful share of considered, researched, high-intent purchases is being mediated by AI, then the unmediated moments (the impulse, the aisle, the scroll, the recommendation to a friend) become disproportionately important to defend. The brands that under-invest in distinctive assets and reach now, on the assumption that AI will do the work for them, are giving up the one terrain where they reliably win.
The second idea is that there is a new form of availability emerging alongside the old one, and it operates by different rules. Call it model availability, machine availability, generative availability — the language clearly hasn’t settled. What it means in practice is the probability that your brand is surfaced, accurately and favourably, when an AI system is asked a question for which your category is the answer.
Building that second kind of availability is not the same job as building the first. It is not primarily about reach, repetition, and distinctive assets. It is about what exists about your brand in the world that the model can read, retrieve, and reason over. Reviews. Comparison content. Independent coverage. Structured data. The accuracy of what is said about you on the surfaces models prefer to draw from. The presence — or absence — of a clear, machine-legible answer to the questions buyers are actually asking.
This is uncomfortable for a profession that has spent two decades getting genuinely good at the first job. The skills, the measurement frameworks, the agency relationships, the briefing rituals — all of them are tuned to building memory structures in human heads. Building memory structures in machine systems is adjacent work, but it is not the same work.
A few things worth sitting with
If you are responsible for marketing or for the capability of marketers in a large organisation, three questions are probably worth a real conversation rather than a quick answer.
What proportion of your category’s buying journeys already start, or end, inside an AI system — and how confident are you in that number? Most marketing teams are guessing, and the guess tends to be low.
How would you know if your brand is being represented accurately in those systems today? If the answer involves someone occasionally typing your brand name into ChatGPT, you do not yet have a measurement loop. You have a hobby.
And, perhaps the harder one: if mental availability and model availability are both going to matter, who in your organisation owns the second one? In most teams, the answer is currently no one — or, more accurately, it is split awkwardly between SEO, PR, content, and the person who happens to be most curious about AI. That is not a structure. That is a gap.
Sharp’s theory is not wrong. It is just no longer the whole picture. The brands that will be remembered — by humans and by machines — are the ones whose marketers figure out how to do both jobs at once.
Forward. Fearless.
For twenty years, Cognitive Union has been helping businesses adapt and thrive. We’ve seen the “box-ticking” era of HR come and go. It doesn’t work. It doesn’t create change.
Making sense of exactly what you are trying to achieve, why you are doing it, and what the result will be. That’s what we bring to the table. We lead from the front, helping you navigate the change, and the “ups and downs” of transformation.
Change is inevitable. But being irrelevant is a choice.
Stop falling into the Discipline Trap. Start building your chains.
Cognitive Union works with large B2C and B2B businesses to build the commercial skills, digital fluency and customer-first thinking their teams need to perform — closing the capability gaps that sit between where teams are today and where the business needs them to be.
Lead. Don’t follow.
Find this blog useful? Sign up to our email newsletter (bottom of this page) where you can receive articles like this and other insights (not publicly published), and you can also follow us on LinkedIn.