Consider Maya, a 36-year-old marketing manager in Chicago who once treated every major purchase like a browser endurance test. She would start by opening Google, Amazon, review sites and several retailer tabs. Now she begins with a prompt for an AI tool. For a recent purchase of a carry-on travel bag, she asked for a lightweight option under $250 that would fit airline limits and come with a warranty. AI narrowed the field and surfaced a specialty brand she hadn’t considered. Maya still checked the seller and shipping terms before buying. The point is that AI had already made her shortlist.
Maya’s new shopping routine captures the latest installment of the PYMNTS Intelligence Consumer AI Benchmark Series. Earlier reports tracked the adoption of AI tools for shopping, users’ experience levels with the technology and how workplace use can lead to personal use. This tenth report turns to commerce. It examines where consumer product research begins, the degree to which shopping is influenced by AI and whether the technology’s suggestions change the price, product, brand or merchant a consumer eventually chooses. AI has become a primary front door to shopping for tens of millions of Americans. Once it owns that first step, it gains influence over the options consumers see and the choices they make.
Four Key Findings
AI has captured the shopping start. About 49.6 million U.S. adults, or 19%, begin their retail product research with AI. Of those, 39 million have stopped using the traditional search channel they once relied on.
AI changes the purchase outcome. Among consumers who used AI to research retail products, 83% report at least one change to their purchasing decision. Forty-three percent found a better price, while 27% chose a different brand and 26% bought a different product.
Both consumers and challenger brands gain new leverage. AI introduced 10.8 million shoppers to a new retailer and pushed 15.9 million toward a different brand. Estimates of savings and merchant gains are directional, but the movement in consumer choice is directly measured.
Ad-supported discovery feels the pressure first. Dedicated AI tools captured 9.5 million retail research starts from Google search and 1.2 million from review sites. That traffic loss threatens models built on page views, ads or affiliate clicks.
AI Owns the First Step
The report’s most important number is 39 million. PYMNTS Intelligence calls these consumers “switchers.” They now mainly start their retail research with AI, not the channel that long functioned as their front door. Roughly 10.6 million more consumers are “augmenters” who use AI alongside traditional channels.
The deeper their AI habit, the more likely a consumer is to switch to AI tools. Power users of the technology abandon traditional starting points at roughly twice the rate of light users. Many begin by adding an assistant, then stop opening the old search box, marketplace or review site as confidence rises.1
Among the 53% of U.S. adult consumers who use the technology for personal tasks, AI touches 28% of their retail purchases on average. Those purchases represent about $74 billion in quarterly AI-assisted spending, based on reported basket sizes. Mainstream users account for $42 billion of that pool because they are a much larger group. That puts the biggest opportunity with everyday users.
The survey measures where research starts and how consumers say AI affected purchases. It doesn’t fully trace where every transaction ends. A shopper may begin with Gemini or Claude, then return to Amazon, traditional search or a merchant site to complete the sale. That distinction will become increasingly important as AI-led discovery moves closer to checkout.
AI Changes What Enters the Cart
Forty-three percent of AI retail researchers found a better price. More than one-quarter chose a different brand, and a similar share bought a different product. Eighteen percent tried a retailer or site they had never used. Fourteen percent bought something unplanned while 19% canceled a planned purchase. Only 17% said the technology had no effect on the outcome.
AI reopens decisions that habit had previously often settled. A consumer who once defaulted to the same brand or marketplace can ask for a recommendation based on price, fit, features, delivery or warranty. Each purchase becomes a fresh contest for a place on the shortlist.
Some of the largest baskets sit with consumers who changed course. Shoppers who made an unplanned purchase spent an average of $1,564 over three months on AI-assisted purchases. Those who tried a new retailer averaged $1,430. Both were above the $931 average among AI retail researchers with purchases. The report shows that consumers are willing to follow AI into new territory they deem commercially valuable.
New Merchants Get a Chance
AI recommendations can give smaller merchants and challenger brands something they have struggled to buy: a place on the shopper’s shortlist. The survey found that 10.8 million consumers who used AI for retail research tried a new retailer or site. Their AI-assisted spending totaled more than $15.1 billion for the quarter and averaged $1,430 per shopper. The report’s model infers that, since most consumers already know the largest marketplaces, many first-time relationships are likely to flow toward smaller or specialized sellers.
That opportunity for brands comes with pressure. AI makes prices easier to compare. A niche merchant may gain discovery while facing a shopper who can test its price, shipping, warranty and return policy against larger rivals. The report estimates $2.2 billion in consumer price savings and $1.8 billion in gross merchandise value shifting toward smaller merchants. Those figures combine survey results with outside benchmarks and assumptions, so they should be presented as directional. A lower product price may also be offset by shipping or other costs.
The Ad Model Loses the Visit
AI’s effect on advertising begins before anyone buys. Search engines, review publishers and comparison websites earn money when a shopper visits a page, sees an ad or follows an affiliate link. When AI provides the comparison inside its own response, the research session may end without producing any of those events.
The survey found that 9.5 million adults who once began retail product research with Google search now start with dedicated AI platforms and assistants. That figure doesn’t include another 10.6 million who moved from conventional Google search to Google’s own AI-powered features. Those consumers changed their behavior, but Google kept them within its ecosystem. Review sites lost 1.2 million former research starters to AI. Across a broader set of research activities, nearly 37% of gen AI users say they visit specialized information sites less often than they once did.
The difference between Google and review publishers is important. The report estimates that about $1.1 billion in quarterly retail search advertising exposure has moved away from traditional search. That calculation estimates the value of research traffic that left search properties rather than documenting an equivalent decline in Google’s reported revenue. Google can also retain shoppers through its own AI products and absorb pressure across a much larger business.
Review publishers have the most to lose as shoppers turn to AI
Review publishers have less protection. Their content may help an AI platform answer a shopper’s question, yet the publisher may receive no visit, ad impression or affiliate click. The report estimates that 18.5% of the audience that once began retail research at review sites has already moved to AI. Review sites face the smallest estimated absolute dollar loss among the affected players, but the largest loss relative to the size of their business model.
That difference makes ad-supported discovery more immediately exposed than marketplace sales. A shopper who starts with AI may still return to Amazon or another large merchant to complete the purchase. For an advertising or affiliate business, however, losing the research start can mean losing the revenue opportunity even when the final transaction happens somewhere familiar.
Implications and Opportunities
For merchants, visibility inside AI-generated answers to consumer queries is becoming as important as placement in search results or marketplace rankings. Product information must be complete, current and easy for AI systems to interpret. Shipping speed, availability, warranty, service and return policies all help an assistant judge value.
For AI companies, product discovery creates rich commercial intent. Their influence remains upstream because shoppers are redirected elsewhere to pay. That leaves an opening for verified merchant information, clearer citations and trusted checkout connections. The survey finds that 61.7% of gen AI users are open to paying inside an assistant, making checkout a likely next point of competition.
Search engines and review sites face different exposure. Google kept another 10.6 million former search-first shoppers inside its own AI features even as 9.5 million left for outside AI tools. Using external ad benchmarks, the report directionally sizes those departures at about $1.1 billion in quarterly retail search ad exposure. Review sites have less room to absorb the loss because AI may summarize their work without delivering the visit, ad impression or affiliate click. The report finds that nearly 37% of gen AI users visit specialized information sites less often.
Maya may still complete her purchase at a familiar retailer. Yet that retailer no longer controls the full path to the cart. AI has influenced which products she considered, which price looked fair and which merchant earned a closer look.
The checkout battle is still developing. The contest for discovery is already well underway.
Read More
PYMNTS Intelligence is the leading provider of information on the consumer trends driving innovation in consumer finance, digital payments and financial inclusion. To stay up to date, subscribe to our newsletters and read our in-depth reports.
Methodology
“The 50 Million Consumer Migration: The Data Behind Retail’s Shift Toward AI Discovery,” the latest installment of the Consumer AI Benchmark Series, is based on a survey of 1,969 U.S. adults conducted in June 2026. The sample was balanced to the U.S. adult population by age, gender, education and income. The commerce module measured where consumers begin retail research, how much shopping involves AI, the value of AI-assisted purchases and the choices AI changes.
Spending was collected in ranges and converted to midpoint values. Outcome groups overlap, so their populations and spending pools shouldn’t be added together. The $74.1 billion quarterly figure comes from weighted survey responses. Estimates involving savings, advertising exposure and merchant gains combine survey data with external benchmarks and should be read as directional.
1. PYMNTS Intelligence segments consumer use of AI tools using an activity-scoring system, not a raw count of tasks. Each AI activity a consumer performs is weighted one to three points according to its complexity and risk. Routine tasks like comparison shopping score low, while higher-stakes tasks like managing personal investments score high.
We define power users as scoring above 27. They represented 7.2% of all consumers in June 2026. Mainstream users, scoring above 4 and up to 27, represented 34.1% of all consumers. Light users, scoring 4 or less, represented 11.9% of all consumers. Holdouts represented 46.8% of all consumers.↩
