Loyalty programs become the identity layer for AI shopping
A quiet repositioning is underway inside retail loyalty programs, and it deserves the attention of every commerce-tech leader. PYMNTS reported on July 20 that purchase histories, stated preferences and engagement signals sitting inside loyalty databases have become among the most actionable first-party data assets a retailer owns. As third-party tracking loses reliability and AI shopping assistants become a primary discovery channel, the programs that companies built for a points economy are being rebuilt for an AI economy. The shift matters because loyalty data now functions as an identity layer, the mechanism that lets an AI agent recognize a customer, personalize a recommendation and connect discovery to verified purchase behavior.
More than six in ten U.S. consumers used a dedicated AI platform such as ChatGPT, Claude or Gemini in the past year, according to the same reporting. That adoption creates a new surface where brands compete for attention, and loyalty data is the credential that carries a known customer across that surface. For leaders weighing where to invest, the message is direct. The loyalty program is turning into core commerce infrastructure, and the data model behind it will determine how well a retailer performs inside agent-led shopping journeys over the next several years.
Ulta shows what a rebuilt loyalty stack looks like
Ulta Beauty offers the clearest working example. The retailer counts more than 46 million loyalty members, and those members account for roughly 95 percent of company sales, a concentration that gives Ulta unusual leverage over its own demand signal. In April the company partnered with Google to embed its assortment inside Google's AI Mode in Search and the Gemini app, placing its products directly in front of shoppers who begin their journey with an assistant. The same loyalty data feeds Ulta AI, a proprietary shopping assistant built on Gemini Enterprise for Customer Experience.
The architecture is worth studying because it runs in two directions at once. Ulta pushes its catalog into third-party AI environments where discovery increasingly happens, and it pulls the same first-party data into an assistant it controls. Both paths depend on a clean, consented loyalty record that ties a person to a purchase history. Retailers without that foundation will struggle to appear intelligently inside external agents, and they will lack the raw material to build a credible assistant of their own. Ulta's 95 percent sales concentration shows how much value flows through a well-governed membership base.
Retail media gains an advantage social platforms cannot match
The loyalty shift also reshapes retail media, one of the fastest-growing margin pools in the industry. Parbinder Dhariwal, vice president and general manager of CVS Media Exchange, framed the advantage plainly. "That relationship plus the loyalty plus the transaction data really sets retail media apart from any other medium," he said. "Not even social media is able to piece all of those constituents together." CVS Media Exchange draws on 90 million customer relationships, a base that links identity, loyalty status and verified transactions in one place. That combination is what advertisers pay a premium to reach.
For commerce-tech leaders, the takeaway is that retail media and loyalty are converging into a single data strategy. The advertising revenue depends on the same identity graph that powers personalization and agentic discovery. Building these as separate systems wastes the asset. We expect the strongest operators to unify loyalty, retail media and AI personalization on shared customer infrastructure, so that one consented profile serves measurement, targeting and recommendation at the same time. Fragmented stacks will leave money on the table and slow every downstream AI initiative.
First-party data turns into an AI training asset
There is a further reason loyalty data has gained value. The signals inside these programs are becoming training and grounding material for AI systems. Companies including Sephora, Fanatics and Ulta are working with providers such as Google Cloud and Eagle Eye to turn membership records into inputs for personalization models and assistants. A loyalty record captures intent, frequency and category affinity in a structured, consented form, which is exactly what a recommendation engine needs to reason about a shopper. That elevates the loyalty database into a strategic data asset for the whole business.
This reframing carries governance weight. If loyalty data feeds AI models and external agents, then consent, data quality and access controls move from compliance concerns to product requirements. A messy or poorly permissioned loyalty database will propagate errors into every assistant it touches. We advise leaders to audit the loyalty data model now, confirm the consent basis for AI use, and treat the membership record as a system of record that other AI services subscribe to. The retailers that get this plumbing right will move faster on every agentic feature that follows.
Personalization is now the baseline expectation
Executives across payments and marketing describe personalization as table stakes, the baseline every competitor must meet. Avery Miller, vice president and head of global loyalty at Visa, and Kipp Johnson, senior director of AI solutions at Braze, both pointed to that baseline expectation, and Jill Moser, senior vice president of offers and loyalty at Mastercard, sits in the same conversation about where loyalty is heading. The common thread is that customers now assume an experience shaped around them, and loyalty data is the fuel that makes such shaping possible at scale.
The competitive consequence is uncomfortable for laggards. When personalization becomes the floor, the retailers still running batch campaigns off stale segments will read as generic inside AI-mediated shopping. Assistants reward the brands that supply rich, structured signals, because those signals let the agent match products to people with confidence. Loyalty programs supply that signal in the highest-quality form available today. Leaders should therefore judge loyalty investment by how well it powers real-time personalization, and give redemption mechanics a secondary place in the business case.
What commerce-tech leaders should do now
We see three decisions worth making this quarter. First, elevate the loyalty database to core infrastructure, with a documented consent basis for AI use and clean identity resolution across channels. Second, decide where the brand's assortment needs to appear inside external assistants such as Google's AI Mode and Gemini, and prepare the product and loyalty data to support that presence. Third, unify loyalty, retail media and personalization on a shared customer profile so the same asset earns advertising revenue and improves recommendations.
The strategic frame is straightforward. Loyalty data has become the currency that AI systems value most, and the retailers holding a rich, well-governed membership base start the agentic era with a durable advantage. Ulta's 46 million members and CVS's 90 million relationships illustrate the scale that turns loyalty into leverage. For everyone else, the work is to convert an existing points program into an identity layer that agents can trust. That project belongs on the technology roadmap today, because the discovery channel it serves is already live.
