After ingesting publisher content, AI companies eye retailers’ vast troves of data

The gaping maw of AI models is always open and hungry for training data. Companies like OpenAI and others have been quick to sign licensing agreements with major publishers, including some of the world’s biggest news organizations.

But to really understand humans, the models need to parse through our behavior — including our shopping preferences.

The retail media industry is full of chatter that networks are being approached to license their data to help inform AI models. At the same time, model developers like OpenAI have started integrating enhanced product recommendations and links to buy into their chatbots. Upon announcing these new features earlier this year, the company explained, “When a user query implies shopping intent, such as ‘gifts for someone who loves cooking’ or ‘best noise-cancelling headphones under $200,’ ChatGPT may surface relevant products.”

And now OpenAI is going even further, showing some brands an early demo of an in-chat checkout feature that would allow users to buy directly within ChatGPT — with the AI firm taking a cut of the sale.

“AI will undoubtedly supercharge the change in shopping experience as LLMs [large language models] move from generic responses to really personalized experiences based on both external training data and data from the user themselves — which is arguably just as powerful,” says Claire Trbovic, group business director, incubation hub at SMG, which runs 10 media networks in connected commerce.

Trbovic says that combining the disparate data sources from publishers, retailers and what users type into their chatbots “could represent the most powerful living dataset known to man — so far.”

The future role of retail media networks

Retail media networks that have the biggest possible base of data to work with will benefit the most, reckons Liam Quinn, director of innovation at e-commerce consultants Visualsoft. “The winners will be those with the largest, broadest and most well-structured datasets,” he explains. “We’ve already seen Shopify partner with ChatGPT to feed their data into the LLM for product recommendations that link directly to a buy button. Amazon is doing similar [things] with their own infrastructure.” Amazon’s AI shopping assistant, Rufus, is forecast to contribute more than $700 million to the retail giant’s operating profits in 2025, according to a leaked internal planning document.

Amazon has also started fending off competitors’ AI agents, including OpenAI, by rewriting its robots.txt file — the document designed to define how automated browsers can interact with a website. Shopify is similarly rewriting its robots.txt file to retain control over the checkout process.

What position retail media networks will play in that future is something that Esme Robinson, platform solutions director at Epsilon, has also been thinking about. “Retail media networks certainly have the data needed to be gatekeepers of consumer intent,” she says, “but will always only represent a small proportion of a shopper’s overall intent and interests.”

Robinson calls them “a valuable tool” that could steal some share from search engines but will be one tool for shoppers among many when researching their potential purchases.

What this means for brands and publishers

In this new world of retail media, Thomas Ives, co-founder of RAAS LAB, an AI-driven scale advertising provider, reckons branding campaigns will need to do a lot of the heavy lifting. “When AI narrows choices to only one or two options, conversion will ultimately come down to brand preference. This loyalty must be secured before the moment of recommendation, making brand advertising more critical than ever.”

Robinson adds that shift in what advertisers actually need to do could be minimal. “It’s always been about the right message at the right time,” she says. “Hyper-targeted product-led ads will be useful at the right point of a shopper journey, but branding will continue to remain important.”

As for publishers, Ives says that this is a boon and not a setback. “They can lean into the value of their trusted, context-rich and immersive environments where brand narratives can inspire discovery and loyalty,” he explains. “Now is the time to champion these branding placements and highlight the deep engagement they offer.”