AI agents are emerging as a new type of customer for major brands, with companies being forced to adapt their marketing and product information for machines that can research, recommend and complete purchases on behalf of consumers.
Tati Lindenberg, global chief marketing officer of Unilever’s homecare division, said retailers now needed to win recognition and trust from both people and automated systems. “The AI answer is becoming a new kind of shelf,” she said.
The shift is already under way. McKinsey research found that 52% of shoppers in the UK had used AI for at least one shopping task in the previous three months. In China, where adoption is more advanced, the figure was 80%.
Most consumers currently use AI to research products before making the final decision themselves. But the role of agents is expected to expand, with Mastercard forecasting that one in 10 people will routinely use them to shop and pay by 2030.
Meta’s personal AI agent, Muse, can research products and make purchases and is expected to launch in the UK “within weeks”. Retailers are already developing ways for customers to buy through services including Microsoft Copilot, Google’s Gemini and ChatGPT.
JD Sports has introduced technology enabling US shoppers to make purchases through those platforms with a single click. John Lewis announced a “significant investment” in AI-powered shopping in March, aiming to ensure its products appear in the same services.
Household products at the forefront
Routine household purchases are among those most likely to be delegated to AI. Nick Carroll, retail principal strategist at Mintel, said consumers were generally most willing to automate categories that were purchased regularly and involved less emotion.
Mintel data showed that 41% of consumers would feel comfortable automating household-product purchases. Beauty, wellbeing and homecare products are therefore expected to be particularly exposed to the development.
The change has become a boardroom priority. Nearly half of consumer-products executives surveyed by EY, 47%, said the ability to influence AI recommendations would be essential to remaining competitive over the next five years. Only 20% believed their businesses could deliver that capability today.
EY has warned that the consequences of failing to adapt could be significant: only 10% of consumers would need to change their behaviour to make a material difference to revenue and profits.
Businesses face a difficult task because AI agents do not operate like conventional search engines. Their algorithms remain poorly understood, while consumers are also asking much more detailed questions.
Fredrik Kridahl, founder of AI search consultancy Hyper Agency, said the average Google search was 3.2 words long, compared with 91 words for an average AI prompt. He also said only 38% of URLs cited by AI ranked in the top 10 of conventional search results, while 31% ranked outside the top 100.
“AI doesn’t care about how big your company is,” Kridahl said. “It cares about the facts and giving the right advice.”
Brands prepare for AI recommendations
Retailers are consequently concentrating on the less glamorous work of making their businesses legible to machines. Sarah Hanna, digital director at the Very Group, said AI search currently accounted for less than 2% of traffic to its website, but that the figure was expected to grow.
“Getting in front of AI agents is set to become just as important as traditional SEO,” she said. Product information must be accurate, detailed and structured so that systems can understand, recommend and compare products, she added.
JD Sports chief technology and transformation officer Jetan Chowk said the company was updating its digital infrastructure so that AI platforms could access current information on products, prices and availability.
Steven Bailey, EY Americas commercial excellence leader, described the process as “unglamorous” marketing to bots. Details including price, promotions, product characteristics, intended customer and points of difference must be encoded in a machine-readable format, he said.
Reviews on retailer websites and social media are also important. A product that is well rated is more likely to give an AI agent confidence when making a recommendation.
Unilever is working on metadata, product pages and online community platforms so that its brands appear in response to natural-language questions. Lindenberg said the company wanted Persil Wonder Wash to be recommended when a shopper asked for a detergent for a quick clean-up.
Human appeal remains important
The rise of machine-led shopping does not mean emotional advertising has become irrelevant. Experts said a brand’s reputation and distinctiveness could make consumers more willing to allow an AI agent to complete a purchase.
Bailey said there was still a place for “romance copy” in the new form of commerce. While AI rewarded information that could be read by machines, it also exposed gaps in a brand’s communication, he said. Trust and humanity could consequently become more valuable points of difference.
Kathryn Gramling, global consumer goods, retail and travel lead at Accenture, said online descriptions needed to convey more than price, size and product features. They should also explain the occasions, preferences and experiences associated with a product, particularly in areas such as fashion, beauty, luxury and travel.
Expedia Group is applying a similar approach to travel. Chief marketing officer Jochen Koedijk said a person might understand the atmosphere of a hotel from a photograph, whereas an AI agent needed that quality translated into evidence such as guest comments and details of amenities.
Lindenberg said brands risked losing their identity if they focused too heavily on algorithms. “The danger is optimizing so hard for algorithms that we lose the creativity and distinctiveness that make a brand worth choosing in the first place,” she said.
She added that as AI made content easier to produce, taste, originality and human imagination would become even stronger competitive advantages.
