What retailers should expect from agentic shoppers this Black Friday
Personal agents aren’t a major channel yet, but the season could offer clues for the future.
• 6 min read
From the way tech companies have talked about agentic commerce, one might expect hordes of digital avatars trampling one another for a TV this Black Friday.
Although the total number of people using AI specifically to take steps toward a purchase remains relatively small, the viral success of Meta’s Muse personal agent and (to a lesser extent) its ilk mean that there will likely be more of these shoppers this holiday season. And retailers and consumer analysts are still figuring out what that means for purchase timing, promotions, and marketing during the biggest week of the year for e-commerce in the US.
Experts told us most shoppers still don’t trust agents to make full purchases on their behalf without a final sign-off. But some might turn to agents to help with shopping tasks, like comparing prices or seeking deals. With Adobe predicting a 95% YoY growth in AI traffic on Black Friday, retailers might treat this holiday season as a test case for early agentic shopping adoption, according to Stephanie Koszyk, VP of the global retail practice at Bain & Company.
“How this all plays out will give us a better indication of some of [these] questions…going into 2027,” Koszyk told us.
Be ready
Koszyk said Bain polling in August found that 24% of consumers said they’d start their shopping journey on an AI platform, up from 17% last year. Another 13% said they would use Amazon or a retailer’s on-site agent.
The most important part of catering to these shoppers and their AI-powered proxies, experts told us, is to ensure all product information is listed in a clear and legible way for crawlers. While Koszyk said the emerging field of generative AI optimization is still evolving, having all of the product data accessible and in order is a baseline step.
“There’s a couple companies that do the research on saying, ‘Across a variety of shopping prompts, how do I show up versus my competitors? Am I one of the top responses or not?’ Koszyk said. “It’s very different than a Google search. So Google search, you could pay for a keyword and show up, and you set it and forget it.”
Brands that have more of their promotional content in video or other multimedia formats may also experience difficulty with AI agents, according to Loni Stark, VP of strategy and product at Adobe. Over time, however, agents will get better at grabbing these formats, as well.
“This is an issue sometimes with the luxury brands we work with,” Stark said. “Because they have spent a lot of time with the videos and everything, so that you can see more of the provenance of how their product was made. What are the ingredients, and why they’re more important. Those things need to be spelled out for the agents, and maybe you didn’t need to do that for humans because they just went for that particular brand.”
Do brands still matter?
One might expect the cold calculations of an AI agent to be less swayed by a glossy logo or flashy brand name. But Koszyk said shoppers can still prompt for specific brand names when tasking their agents to find a product.
Stark said agents might also pull from Reddit threads or product discussions for information on the quality of specific brands. Or they might associate certain brands with better overall delivery or return policies.
Retail Brew delivers the latest retail industry news and insights surrounding marketing, DTC, and e-commerce to keep leaders and decision-makers up to date.
By subscribing, you accept our Terms & Privacy Policy.
Jennifer Howard, chief merchant at digital commerce security company Forter, said personal agents might be already primed on the brands that their users prefer. “How do you get people to tell the agent, ‘Look here first’?” Howard said.
Even still, Koszyk said AI recommendations do tend to favor “insurgent brands”—smaller, fast-growing companies with devoted online followings.
“The way that [LLMs] come up with their responses, they are going through retailer brand websites to get information, but they’re pairing that with reviews on Reddit and other community forums, and so what they’re really looking for is consistency and convergence,” she said, adding that these brands “have something…that’s bubbling up from the consumer and the shopper experience, and so that’s why they tend to do better than technically the market share leaders in a lot of these categories.”
Robo promo
Howard said retailers should pay attention to the timing of how agents access Black Friday deals. Retailers tend to run new promotions each day from Black Friday to Cyber Monday and even Small Business Tuesday, she said. But in a world where agents could be monitoring discounts, they might access pages with different frequency.
“The question that I would ask myself is: Is that the best way to run my promotions, given that agents are going to be doing the scraping?” Howard said. “How sophisticated are the prompts within the agentic commerce space? And then how sophisticated are the answers?”
It will be important for retailers to watch if agents are able to help consumers take better advantage of promotions offered during the post-Thanksgiving sales holidays and potentially adjust their discounting in the future, Koszyk said.
“Especially for the Black Friday and Cyber Monday weekend, [what’s] important is understanding at what price point most of the volume is coming through,” Koszyk said. “Did the agents actually help people buy more on promotion? Are they using it that way in terms of getting the best deal because it’s always on? Versus a human would have to periodically go and check.”
To block or not to block
There’s also the question of whether retailers want outside agents sifting through their sites in the first place. Amazon blocked Muse from its platform while also offering its own shopping agent. Other retailers offer similar on-platform agentic experiences. Does it make sense to allow Muse or other agents to collect data and potentially botch orders?
Koszyk said most retailers don’t have the Amazon-level clout that could afford them the freedom to block third-party agents. Consumers do tend to trust agents hosted on retailer websites more than generic LLMs, she said, but that gap has narrowed since last year.
“[AI platforms are] where there is usage and growing usage over the past couple years of where people want to start [shopping],” Koszyk said. “There is still an opportunity for people to do it with their own [first-party] agents, and a lot of them are doing both and should be doing both.”
Howard said each approach might depend on the size of the retailer and its specific customer base, however.
“There’s no one size fits all. You have your Amazon and Walmarts—that can’t compare even to a company like Gap, which is a huge company, but not at the scale of Walmart and Amazon,” Howard said. “Everybody is at a different stage…Their customers are different. Every merchant needs to make the decision for themselves.”
Retail Brew delivers the latest retail industry news and insights surrounding marketing, DTC, and e-commerce to keep leaders and decision-makers up to date.
By subscribing, you accept our Terms & Privacy Policy.