Much of the oxygen surrounding AI within marketing is focused on automation, efficiency and personalisation. These are useful conversations in their own right, but they sit downstream of a bigger conversation: whether the systems now shaping discovery can correctly understand what a brand is, where it belongs and why it should be recommended.
Marketers and business leaders need to understand how the customer journey has shifted over the last 12-18 months. Discovery, once shaped by people actively searching, browsing or comparing, is now shaped by systems that classify, rank, summarise and recommend before a customer has made a conscious choice. Search engines, social platforms, marketplaces, comparison sites, retail media networks and AI assistants all now play a role in deciding which brands become visible, which are trusted, and which are ignored.
The real-world implication is that a brand can spend years building a position in the market, only to find that the systems mediating choice understand it in a thinner, weaker or less accurate way.
In practice, this might look like a brand being described as a generic option when its value lies in specialism. Or it might be grouped with the wrong competitors. It might fail to appear for the category questions it should naturally own. It might be surfaced to the wrong type of customer because platform learning is being fed by low-quality signals. These are fundamental discovery problems, and discovery problems become growth problems.
Research across 197 brands in Fashion, Finance, General Retail and Travel shows that discovery is not a vague brand concept. It is a measurable growth system, and the brands investing in it properly are pulling ahead.
The important point is that discovery does not work in the same way everywhere. Different categories reward different forms of discoverability. In Travel, being found carries more weight because people often begin with a destination, date, route or budget rather than a fixed brand preference. In Fashion, the work is more balanced because people are interpreting taste, identity, quality and social meaning. In Finance, buyers look for risk-reducing cues: recognition, authority, reviews, ease and enough proof to feel safe. In General Retail, the job is often more task-based, where stock, delivery, price, reviews and product detail may matter more than a broader brand narrative.
AI makes this category difference more important, because machine interpretation is now part of the discovery system. A brand has to make sense to people and be legible to the systems helping those people decide.
That creates two layers of interpretation.
The first is human interpretation: what people remember, recognise, trust, say, feel and repeat. This is the work marketers are used to managing, even if they do not always measure it well enough.
The second is machine interpretation: how search engines categorise the brand, how LLMs summarise it, how marketplaces assign relevance, how ad platforms learn who is likely to respond, and how recommendation systems decide where the brand belongs.
The growth risk is that these two layers can drift apart. A brand may be well understood by its existing customers but poorly represented by AI assistants. It may be visible in search but attached to the wrong category language. It may perform well in platform dashboards but return diluted lead quality or low-value customers. It may have strong creative, but weak structured signals. Each gap creates leakage.
This is why AI visibility alone is too blunt a goal. The question can start with whether the brand appears in an AI-generated answer, but it needs to expand to examine whether the brand appears in the right context, with the right associations, for the right customer need, in a category where that behaviour actually affects choice.
That last point matters more than most AI strategies acknowledge. LLM visibility is already helping in fashion and travel industries, where people use AI tools for inspiration and shortlisting. In finance and general retail, the relationship is not yet translating into demand in the same way. In those categories, people are not switching banks or buying sofas simply because an AI tool mentioned a brand. Marketing must flex their approach to accommodate different consumer behaviour patterns.
This is where many AI strategies become too generic. They assume that appearing in AI answers is always a win. A more effective strategy is to understand how discovery works in your category, then decide what role machine visibility should play.
There is another caution in the data. Across every sector in the research, human signals still carry more predictive weight than machine signals. The dynamic varies by sector, but algorithmic discovery channels never deliver more impact than what real people remember and repeat. Machine systems amplify signals; they are much less reliable when asked to compensate for weak meaning. A brand that is not remembered, trusted or talked about gives machines less to classify with confidence. The system can surface content, but it cannot create a strong brand interpretation from weak evidence.
In 2026, the ultimate job for marketers is to make those two layers work together. Human understanding gives the brand meaning, and machine understanding helps disseminate that meaning to more humans.
When the layers work together — and when they don’t
When the two align, demand can compound. People search for the brand using the right language, AI systems describe it accurately, reviews reinforce the right proof points, paid platforms learn from higher-quality actions, marketplaces provide better context for products, and search engines connect the brand to the right category need. In essence, the brand becomes easier to find because it has become easier to understand.
When the two diverge, the symptoms are harder to read. Growth slows without dashboard data that identifies the cause. Media costs rise. Traffic comes in, but quality falls. AI mentions increase, but demand does not move. Search visibility improves, but the brand is still not being chosen. The business sees performance friction – at the root, the issue is interpretation.
That is why more visibility cannot always fix the problem. If the brand is absent from the right places, more presence may be exactly what is needed. If the brand is being misread, more presence may simply scale the misreading. The system gets more data, but the data can absolutely deliver the wrong lesson.
Audit before you accelerate
Marketers need to conduct a five-step audit to understand their growth blockers before they begin designing a solution.
- The first audit is category reality. What kind of discovery does the market reward? A ‘found’ gap in the travel sector is not the same gap in finance. A ‘chosen’ strength in retail does not mean the same thing in fashion. The diagnosis has to be read through how people actually buy in that market.
- The second audit is human interpretation. What do people remember? What do they repeat? What do they trust? What language do they use? What proof do they need? What keeps them from choosing?
- The third audit is machine interpretation. How do search engines categorise the brand? How do LLMs describe it? Which competitors appear alongside it? Which attributes are repeated? Which customer needs trigger the brand to appear? What do ad platforms seem to be learning from conversion signals? How is product data being read by marketplaces?
- The fourth audit is the gap between the two. This is usually where the richest insight sits. A brand may be loved by customers but absent from AI answers. It may be well represented in organic search but misunderstood by paid platforms. It may be known for one thing by people and another by machines. It may be strong in reviews but weak in the category language that discovery systems rely on. Each gap points to a different growth blocker.
- The fifth audit is momentum. Once the constraint is clear, the business needs to know what to do next. The answer may be better content structure, clearer category language, stronger proof points, improved feed data, review volume, refreshed creative, cleaner conversion signals, a different measurement approach, or a sharper role for paid media. The intervention should match the constraint
The real competitive advantage in the AI era
The brands that win the next phase of AI will be effective in driving visibility and the easiest to understand, easiest to classify, easiest to trust and easiest to choose. That takes more than technical optimisation; it takes a clear view of how the brand is interpreted by humans and machines, where those interpretations diverge, and what needs to change to close the gap.
