Artificial intelligence’s dramatic upending of search and discovery the last few years still hasn’t transformed most consumers’ shopping experience. But given large language models’ (LLMs’) rapid adoption over the past two years, this could change fairly quickly.
Currently nearly half of consumers say AI hasn't changed their shopping experience, according to new data from Attain. About 49% report that their shopping feels exactly the same as before. Meanwhile, only 18% of shoppers feel recommendations have become more personalized, and 33% remain entirely uninterested in AI shopping tools.
And just what is the missing feature most shoppers desire? Simple price drop alerts, which was sought by 36% of consumers Attain surveyed.
This disconnect between AI’s promise and consumer reality reveals a missing ingredient: a deeper, more holistic understanding of consumer purchase behavior that would allow for greater personalization. The transformative AI revolution promised by tech evangelists, aligns with coupons and holiday sales.
The Attain numbers paint a picture of tentative adoption and unfulfilled potential.
In particular, the results seem to ask: what if AI agents could learn not just from internet scraping, but from actual purchase patterns? This shift from theoretical to behavioral data could fundamentally alter how artificial intelligence serves consumers and brands alike.
In that sense, the most immediate change AI is bringing to the way retail brands and products meet consumers is likely to be felt through ad targeting.
“The inclusion of transaction data will enable more precise targeting, real-time attribution, and measurable outcomes, bridging the gap between media investment and business impact,” says Dan Kurtter, SVP of Strategy and Product at Attain. “Agents will use this to create more personalized commerce.”
On the flip side, consumers are gravitating toward more practical applications. Beyond price alerts, 23% want help comparing products and reviews, while only 7% desire an AI assistant that can place orders autonomously, Attain finds. This pragmatic approach suggests consumers want AI to enhance rather than replace their shopping decisions.
Stefan Hajek, executive creative director at design and branding firm Designit, argues for a fundamental shift in perspective.
“Marketers need to treat purchase data differently. It can no longer be viewed as a receipt at the end of the funnel,” he says. “Rather than using it to analyze the past, it should be viewed as a narrative blueprint for the future. This data is intent-predictive, facilitating real-time personalization and smarter campaign creation and optimization.”
The integration of permissioned purchase data into AI models represents more than incremental improvement. It's a major shift from guessing to knowing, Hajek suggests.
When language models can access real transaction histories, they move beyond pattern matching to understanding actual consumer behavior. This creates possibilities for hyper-personalized recommendations that reflect not just browsing history, but purchasing power, brand loyalty, and seasonal patterns.
Of course, privacy considerations and legislation loom large in this evolution. The success of purchase-data-driven AI could depend on transparent data practices and consumer consent. Brands that can navigate these ethical waters while delivering genuine value stand to gain significant competitive advantages.
“Purchase data can help make content creation far more outcome-driven,” Hajek notes. “If models can be trained to iterate based on behavior rather than prompts, then marketing can become more dynamically personalized, and rather than targeting personas, content can be designed and optimized based on behaviors.”
The Attain data reveals an adoption curve still in its infancy — 22% use AI tools regularly for shopping, while 27% have tried them only once or twice. This suggests enormous room for growth as models become more sophisticated and useful.
Kurtter envisions an even broader transformation.
“Software will grow more intuitive and interactive, lowering the barrier to entry for non-technical users,” Kurtter says. “Intelligent agents will drive interoperability across platforms, breaking down silos.”
As AI models begin incorporating purchase data, the industry faces a critical inflection point. The challenge isn't just technical—it’s about creating experiences that consumers actually want.
The modest enthusiasm for current AI shopping tools suggests that true transformation requires more than better algorithms; it demands understanding what moves people from browsing to buying.
“When LLMs can learn from actual buying behaviors, then they can help us to understand the tangible impact,” Hajek says. “We move from measuring impressions to outcomes. And the beauty of this is that creativity gets braver when we're not guessing what might work - rather, we're realizing what does.”
