Beyond search: Making ‘Share of Model’ work for your brand
Opinion – Week in Focus
Share of Model helps measure how often AI mentions and describes a brand, product, or attribute, and in what terms. Jellyfish’s senior media director offers advice on how to use it to rethink search.
Search used to be the default way to understand how a brand was discovered. That’s changing fast. Today, people are asking AI models what to buy, which brands to trust, and which products best fit their needs.
Share of Model helps measure how often AI models mention and describe a brand, product, or attribute, and in what terms. So, the question is no longer just where a brand ranks, but how it is evaluated, selected and described when the answer is generated.
In this context, visibility isn’t the goal; it’s the result. Being cited by AI reflects how well a brand or product aligns with the criteria models use to compare and recommend options across real user queries.
Treat the model like part of the market
The starting point for leveraging Share of Model is fairly simple. If LLMs are shaping how people research products, marketers need to understand how those models talk about brands in the first place.
By looking across models, prompts, and categories, marketers can see which attributes keep appearing, which products are surfaced, and which strengths or weaknesses are attached to a brand.
But there’s a wider shift behind that. Share of Model provides a way to continuously measure brand perception across multiple AI systems. The model is becoming part of the route people take to compare, shortlist, and ultimately choose, giving its recommendations an influence media teams can’t ignore.
In AI-driven journeys, what matters operationally is not just being visible, but being selected, trusted and recommended at key decision moments.
Use the signal, but stay in control
The value here lies in the signal, not in surrendering judgment. AI models can hallucinate, over-index, or suggest themes that conflict with brand goals – so workflows must be designed to catch that.
That’s why the process is more considered than the phrase ‘AI optimisation’ sometimes implies.
Recommendations are shaped by brand guidelines, refined through additional checks and reviewed by human practitioners before anything is applied. A model might describe a luxury brand in terms of affordability, but that doesn’t mean those messages should be used in a live campaign.
The human-in-the-loop point matters for another reason, too. Model perception is influenced by what people say, write, and search for, but those same people are increasingly influenced by the models in return.
That can be a self-perpetuating cycle, making it critical to observe that layer of perception and weigh it against what a brand already knows about its customers, category, and voice.
Connecting AI visibility with media activation
As AI systems become increasingly embedded in decision-making, a new opportunity is emerging for media teams: linking Share of Model insights to paid media activation.
On one side, GEO monitoring provides a continuous view of how models perceive brands, products, and attributes. On the other hand, advertising is increasingly appearing directly within AI environments, while traditional platforms still capture demand shaped upstream by these systems.
Bringing these together allows teams to move from observation to orchestration – prioritising investment based on AI-driven demand signals, aligning messaging with how models frame the category, and activating campaigns to reinforce or rebalance brand influence where it matters most.
Watch the differences, not just the average
“AI” is not one thing. Different models rely on different sources, produce different answers, and can shape very different impressions of the same brand. Tracking these source patterns over time reveals how brand perceptions shift constantly, and from model to model.
This matters both operationally and strategically. A brand might be well represented in one model’s answers but much less visible in another.
If a model consistently leans on a specific publisher, marketplace, or review content, it can influence which claims, features, or product angles rise to the surface.
From keywords to semantics
One implication of this shift is how marketing strategies are structured. Optimisation is moving away from isolated keywords towards a broader management of semantic territories.
Rather than focusing solely on which keywords to bid on, teams need to understand which themes, attributes, and use cases define the category, and where the brand is strongly represented, underrepresented, or absent.
These gaps can then be addressed through a combination of GEO, AI-native paid activations, and traditional media optimisation, informed by enriched signals such as synthetic consumer panels.
This approach allows brands to capture demand and also anticipate and shape it earlier in the decision journey.
The future of search
Answer engines are moving beyond discovery and comparison towards transaction, which raises the stakes as Share of Model evolves from research insight to a fundamental part of commercial reality.
Previously, visibility was a ranking problem. Now, it is a perception opportunity: generated answers don’t just point people somewhere; they frame the brand before the click, and sometimes before a click even exists.
Search still matters – but it’s no longer the only place where discovery, persuasion and commercial intent are shaped.
Victor Batista is the senior media director of Jellyfish