Digital advertising’s potency has always come down to knowing what the audience is searching for and the sites they’re visiting online.
But what if your audience isn’t looking at search results anymore? What if they’re seeking recommendations from ChatGPT instead?
According to Jellyfish, an ad agency that specializes in digital platform advertising, this isn’t just a hypothetical scenario. Research that the Brandtech Group-housed agency conducted with YouGov suggests that two-thirds of 18-24-year-olds now use AI models for recommendations, and half of them expect AI tools to make the best choices.
To solve this AI problem, Jellyfish is proposing an AI solution: a Share-of-Model platform, which was officially launched in early access last month.
The product is designed to work the same way a typical market research survey does, except the respondents are popular large language models (LLMs) like Anthropic’s Claude, OpenAI’s ChatGPT, Meta’s Llama and Google’s Gemini.
“In some cases, they’re arguably the most important members of your audience,” said Jack Smyth, Jellyfish’s chief solutions officer for AI. “How they define a category will shape real-world activity.”
I heard what you said
Each day, the Share-of-Model product sends queries to these LLMs about different categories, brands, products and audiences, such as “What brand of X would you recommend?” and “What do you think about Brand Y?”
Responses get collected into Jellyfish’s own AI model for pattern analysis and clustered into different searchable categories based on keyword volume, sentiment and model source.
As a brief demonstration, Smyth presented a chart that a telecommunications company interested in brand perception might find useful, which depicted brands in the category that are most positively associated with network coverage, customer support and value.
Jellyfish’s new AI query-collector can also track how each individual LLM responds to questions about a brand; Claude 3.5 might be more likely to bring up positive reviews, for example, while GPT 4o focuses on sustainability efforts.
French food and beverage company Danone (which sells many of its products under the name “Dannon” in the US) was Jellyfish’s first official client. Catherine Lautier, Danone’s VP and global head of media and integrated brand communication, told AdExchanger she was excited by the platform’s potential after previewing the technology at Cannes last June.
“It’s helping us understand competition from a digital brand ecosystem point of view, but also how we can build meaningful and different shaded brands on certain attributes within a category,” said Lautier.
Changing algorithms over minds
But what can a brand actually do to influence an LLM’s responses?
So far, the winning method seems to be what works for search engine optimization (SEO): Create more straightforward, educational content for the brand, and make sure it’s easily accessible.
In fact, Jellyfish has already begun testing multimodal AI models that analyze images, video and podcast audio to provide feedback to brands.
“We understand how the LLMs are scrolling our websites and gathering information,” said Lautier. She added, for instance, that Danone is particularly focused on uplifting how generative AI models refer to its yogurt and baby nutrition brands.
However, the Share-of-Model platform can’t stop LLMs from “hallucinating” – the cheeky, personified term for when the generated response is completely false or made-up.
“The job of the Share-of-Model is not necessarily to get the most accurate information from the model – it’s to help a marketer understand what the model might be saying to customers,” said Smyth. “Sometimes that delta is an insight in its own right.”
One could argue it’s similar to word-of-mouth marketing in that way. A brand can’t stop a person from incorrectly recalling a product’s benefits in conversation. It can only continue to focus on educating its customers – and hope the person nails the pitch anyway.
Oops! All Bots
Ironically, one of the biggest criticisms with SEO is that companies end up prioritizing how a web crawler responds to a site, rather than an actual human. That seems like an obvious concern with LLMs, too – that brands will focus even more of their attention on marketing to bots and algorithms.
But Smyth finds the idea of creating content for AI agents (particularly multi-modal agents) more “liberating for a marketer,” specifically because “you’re moving beyond the constraints of how much you can pack into text.”
“Think of what it means for packaging, for logos, for the entire output of our industry,” Smyth said.
All this could sound a bit dystopian for someone who doesn’t yet use AI to grocery shop. But Smyth made it clear that the goal is to “build a differentiated, meaningful brand to two different types of intelligence” – which, thankfully, still includes people.
In the meantime, although AI adoption is climbing, it’s not yet so rampant that brands are rushing to pivot. Another recent YouGov survey reports that two in five Americans think of generative AI more negatively than they did a year ago, and still another suggests that 46% would never seek help from a chatbot to make a decision.
So, maybe don’t convert all your nutrition facts into binary just yet. A lot of us still like to read the labels.
