Ulta is marketing to AI agents and shoppers

A shopper types "skincare for acne" into ChatGPT or Gemini and gets an answer at the top of the prompt result. Kelly Mahoney, chief marketing officer at Ulta Beauty, says the retailer is now working to earn that spot. That means creating content and experiences for a new audience. "We say we want to be the best marketers to our consumer, but we also need to be the best marketers to the agent," she said at the Status Summit.

This goes beyond search rankings. Mahoney said consumers are using AI tools "to not only discover, but now and more and more to also make the purchase." If AI agents become the first place shoppers ask where to buy, a retailer's marketing has to convince software as well as people. Mahoney's view is that beauty marketers have handled this kind of problem before. Her answer combines first-party data, managed relationships and physical stores.

Agents are the newest voices a brand can't control

Mahoney doesn't treat AI-driven discovery as completely new. In beauty, she argued, persuasion already happens mostly through other people. "Gone are the days where you can put out an ad and then just sit back," she said. Ulta still runs traditional advertising, but she described creator-led marketing as the main way customers are influenced. The brand gives up control over how it is described, while staying clear about "who you are, what you represent, how you want to show up."

In her account, agents fit the same model. "It's just that it's agents. It's non humans," she said. A creator decides whether to recommend a product. An AI agent decides which content to surface when a customer describes a problem. Either way, the brand doesn't write the recommendation. It can only shape what the recommender knows.

So Ulta's questions are practical ones: what do the agents want to know, how should Ulta supply it, and what content would lead an agent to decide that "Ulta Beauty is the place you go to get what you need."

Ulta already does this with two groups of people. The first is store associates, whom Mahoney calls the brand's ambassadors: "who better but the associates in your store to talk about what makes Ulta Beauty so special." The second is the Ulta Beauty Collective, a group of influencers. Both relationships involve regular meetings, briefings and events so the people speaking for the brand know what to say. The next step, she said, is to apply "the same concept just now for AI and these agents."

The loyalty data underneath it all

Ulta's starting point is its loyalty data. Mahoney has been with Ulta Beauty for eleven years, and she recalled that when she stepped into the role, the Ulta Beauty Rewards program had about 11 million members. She described today's members as engaged and active, and said the business runs on them.

With nearly all sales tied to identified members, Ulta has what Mahoney calls a rich first-party data infrastructure. She said the company built it about ten years ago and has rebuilt it "a couple of times" to keep the data ready to use. She called personalization an overused word. For Ulta, it starts with something else: resetting segmentation around customer motivations instead of basics like gender and demographics.

Those segments draw on non-purchase signals as well as transactions: social media activity, how customers interact with Ulta, whether they come into stores, and whether they leave ratings or reviews. Only after that does Ulta apply personalization tactics such as recommendations. Her example was a customer known to shop in stores, who might get a message in the right channel saying hours have been extended in their region.

Mahoney didn't draw a direct line from this data to Ulta's agent work. A reasonable reading, though, is that a retailer that already knows which motivations and problems bring customers in is better placed to decide what content agents need.

No playbook, so measure everything

Mahoney was open that nobody has a proven method yet. "I don't think there's a playbook, so it's really stepping through this. But you can't wait. It's here," she said. She described a quick change at the executive table, from leaders wondering how much to invest before they understood AI to a shared view that companies "better be all in."

Her answer to working without a playbook is measurement. She spent ten years in finance at BP before moving into marketing, and she credits that analytical background with building her performance marketing skills: loyalty, data, personalization, measurement and customer lifetime value. She said that maybe five years ago she wasn't sure she would ever become a CMO, because she hadn't come up on the traditional brand side. She argued the job has since moved toward her skill set.

On agents, she said, "all of this is measurable." Teams can test, learn and adjust their content and platform choices without waiting for a standard approach to appear.

Stores are where agent picks get checked

A strategy built around agents might seem to make stores matter less. Mahoney argued the reverse, based on Ulta's research on Generation Alpha, which the company calls the next generation. She said the research shows these consumers are highly educated about beauty and engaged with it at younger ages, mainly through skincare rather than makeup.

She gave an example from home. At a sleepover for her daughter, who is 11, two of the four guests forgot their pajamas but brought every skincare product possible in their bags.

The finding she called "kind of a sigh of relief" was that young consumers who use AI tools and chatbots "actually crave coming into the store" and are more likely to visit. The interviewer described the same pattern: younger shoppers find inspiration on YouTube and Pinterest, then come into stores to confirm what they found.

Mahoney said this puts more pressure on the store to earn the trip. Ulta is investing in how its stores look and in immersive events, and it has run live selling inside stores through its TikTok Shop integration. She called that an example of "the two worlds" blending.

Asked to finish the sentence "The future of brand discovery will be...", Mahoney said it will be "very, very fluid," and that brands must pair "everything you're doing digitally and through agents with real life." In that model, an agent can send a shopper to Ulta. The store is what keeps them coming back.