AI Shopping Agents Are Already Picking Retail's Winners

A single test of 177 toy brands just exposed one of the biggest blind spots in modern marketing. When Amazon's Alexa for Shopping was asked for toy recommendations, it pulled almost every single result from Amazon's own store. Alexa for Shopping showed toys from 177 brands when asked for recommendations and nearly every one of them came from Amazon's own store. Ask ChatGPT the identical question and it spreads its answers across 24 retailers . Ask Google's AI Mode and it widens the field to 37 different retail sources.

That is not a rounding error. It is proof that each AI shopping agent runs on its own invisible rulebook, and those rules are already deciding which brands get seen and which ones disappear. Matt Britton, founder of consumer intelligence platform Suzy and author of the upcoming book Generation AI, has spent the last year warning Fortune 500 marketing teams that this shift is happening faster than their measurement systems can track.

AI shopping agents are software programs, embedded in tools like ChatGPT, Google AI Mode, and Amazon's Alexa for Shopping, that research, compare, and recommend products to consumers in natural language rather than a list of search links. They are quickly becoming the new front door to retail. AI now plays a role in 86% of shoppers' retail journeys , and among general-purpose AI tools specifically, 68% of more than 3,000 U.S. consumers surveyed used a general AI chatbot for at least one shopping task in the past three months .

The stakes go beyond traffic. 62% have purchased a different product or brand instead of their usual choice because of an AI chatbot . That means every brand excluded from an AI assistant's shortlist is not just losing a click, it is losing the sale to a direct competitor the algorithm decided to surface instead.

Britton calls this emerging discipline "share of model," the AI-era equivalent of share of shelf. Where retail strategists once fought for end-cap placement and premium shelf positioning, they now need to understand how often, and how favorably, each major AI assistant mentions their brand when a shopper asks for a recommendation. This post breaks down what the new data reveals, why the rules differ wildly by platform, and what marketing leaders need to do in the next two quarters to protect market share nobody has fully learned to measure yet.

What Is "Share of Model" and Why It Rivals Share of Shelf

Share of model measures how consistently an AI shopping agent recommends a specific brand or retailer across relevant customer queries. It is the direct descendant of share of shelf, the decades-old retail metric that tracked physical shelf space as a proxy for sales potential. The difference is that shelf space was visible and negotiable through buyer meetings and slotting fees, while share of model is governed by data feeds, training sources, and proprietary ranking logic that most brands cannot see or influence directly.

The toy aisle test illustrates exactly how much this varies by platform. AI visibility hinges on a brand being stocked in the stores each assistant is willing to look at, and which retailer lands the sale depends largely on which assistant the shopper happened to open. A brand sold exclusively through independent retailers could be functionally invisible on Alexa for Shopping while performing well on ChatGPT or Google AI Mode, simply because of where each assistant chooses to look.

Microsoft's Copilot behaves differently still. Of all the assistants, Microsoft's Copilot cites retailers the most, according to Tinuiti, which tracked which e-commerce sites the assistants cited most often in answers to its prompts. That means a brand's visibility strategy cannot be built around a single AI platform. It has to account for at least four or five distinct gatekeepers, each with its own appetite for retailer diversity.

Britton argues this fragmentation is exactly why most CMOs are flying blind right now. Boards have spent a decade building dashboards for paid search, social, and marketplace share. Almost none of them have a line item for how their brand performs inside a ChatGPT conversation or an Alexa for Shopping query, even though those conversations are already redirecting purchase decisions at scale.

How Agentic Commerce Is Redirecting Purchase Decisions Today

Agentic commerce refers to the growing category of AI tools that do more than recommend products, they actively compare, shortlist, and in some cases complete purchases on a consumer's behalf. Adoption is accelerating well past early-adopter territory. In Attentive's survey of 3,054 U.S. adults conducted in August 2026, 68% had used a general AI chatbot for at least one shopping task in the prior three months.

Retailer-owned assistants in particular are converting that interest into real revenue. Amazon says shoppers who use Rufus are 60% more likely to finish a purchase, and it expects the tool to generate more than $10 billion in yearly sales. That single data point should reframe how every retail and CPG marketing team thinks about platform-specific AI optimization, because the financial upside for being recommended well is no longer theoretical.

Clothing and footwear are emerging as one of the categories most open to full agentic handoff. Clothing and footwear in particular were among the retail categories where shoppers would be the most willing to hand over spending power, next to groceries, with 69 percent of respondents surveyed saying they would let an AI agent spend up to 100 US dollars on clothing and footwear. Shoppers are not just asking AI for advice anymore. They are increasingly comfortable letting it transact, provided the dollar amount stays modest and reversible.

Still, trust has clear limits that brands need to respect in how they design the AI experience. 53% are uncomfortable allowing an AI assistant to make purchases on their behalf, 20% would accept recommendations from AI but not autonomous purchases, 14% would require manual approval for every purchase, and only 7% would permit autonomous purchasing under predefined conditions. Britton points out that this gap between willingness to be influenced and willingness to fully delegate is where the real opportunity sits for brands willing to build transparent, controllable AI shopping experiences rather than fully automated black boxes.

Why Brands Still Shape the Journey, Even When AI Leads Discovery

The data makes clear that AI influence does not eliminate human judgment, it reshapes where that judgment gets applied. 71% say AI meaningfully affected which products or brands they seriously considered or chose at least half the times they used it to shop, and 93% sought information or proof outside of AI while deciding whether to purchase during a recent AI-assisted shopping experience. Consumers are treating AI recommendations as a shortlist generator, not a final verdict.

That validation behavior matters enormously for where brands invest their content and proof points. Shoppers who double-check an AI recommendation are most often most often customer reviews (37%), pricing across retailers (36%) and detailed product information (33%) . A brand that wins the AI mention but has thin reviews or inconsistent pricing across channels can still lose the sale in that validation step.

Checkout location preference also remains firmly brand-controlled for now. 60% prefer to complete a future purchase on the website or app where the product is sold. That is a meaningful signal for retail and e-commerce leaders: owning the AI mention is only half the battle, the post-recommendation experience on the brand's own site or app still closes the majority of transactions.

Britton uses this exact finding in his keynote work with Fortune 500 clients to push back on the panic narrative that AI will simply disintermediate every brand relationship. Instead, he frames it as a redistribution of where trust-building has to happen. Brands that treated their website as a static storefront now need to treat it as the final, decisive stage of an AI-initiated shopping journey.

Where AI Actually Sources Its Shopping Recommendations

Perhaps the most uncomfortable data point for brand marketers involves where AI assistants pull their information from in the first place. Across a large sample of purchase-intent queries tested on Google AI Mode, ChatGPT, and Perplexity, only 2.8% of the 1,851 sources cited across Google AI Mode, ChatGPT and Perplexity were brand-owned pages. The largest single source category was something else entirely. The largest source category made up 59% of citations, handing AI visibility to third-party publishers instead of brands.

This means the traditional SEO playbook, built around optimizing the brand's own domain, is necessary but no longer sufficient. Brands now have to earn visibility on the review sites, forums, and publisher content that AI models treat as trustworthy sources. That is a fundamentally different discipline than classic on-site SEO, and most marketing organizations do not yet have a team or budget line assigned to it.

Agreement between models is also far lower than most executives assume, which compounds the difficulty of any single-platform strategy. One large-scale audit running 100 real shopping questions through ChatGPT and Google AI Mode, auditing all 138 products they recommended, found the two agreed only 38% of the time. That level of disagreement means a brand cannot assume that winning on one assistant translates to winning on another.

Britton's recommendation to clients is straightforward: treat AI visibility auditing the way brands once treated search engine rank tracking, except across five or six platforms simultaneously instead of one. His team at Suzy has built consumer intelligence tools specifically to help brands understand how real shoppers are interacting with these assistants, closing the gap between what a brand assumes AI is saying about them and what it is actually saying.

Building an AI Visibility Audit for 2027 Planning Cycles

Marketing leaders heading into 2027 budget planning need a structured way to assess exposure across the major AI shopping surfaces. The following checklist reflects the framework Britton walks Fortune 500 teams through during his AI keynote engagements:

• Map category-specific recommendation behavior on Alexa for Shopping, ChatGPT, Google AI Mode, Perplexity, and Copilot for your top five product categories, since retailer diversity varies dramatically by platform.
• Identify which third-party sources each assistant cites most for your category and build a relationship or content strategy around those specific publishers and review sites.
• Audit pricing and product data consistency across every retail partner, since shoppers cross-check pricing before completing nearly four in ten AI-influenced purchases.
• Strengthen the post-click experience on owned websites and apps, since most AI-assisted shoppers still prefer to complete the purchase there rather than inside the chatbot.
• Establish a recurring reporting cadence for share of model alongside existing share of shelf and share of voice metrics, so leadership sees AI visibility trends before competitors capitalize on gaps.

Companies in regulated or high-consideration categories face an even sharper version of this challenge. Industries like financial services and

Share of model is a term describing how consistently and favorably an AI assistant recommends a specific brand across relevant shopping queries. It functions as the AI-era equivalent of share of shelf, except the rules governing visibility differ by platform, with data showing that Amazon's assistant, ChatGPT, and Google AI Mode each surface dramatically different sets of retailers for identical product requests.

Do AI shopping recommendations actually change what people buy?

Yes. New research found that 62% have purchased a different product or brand instead of their usual choice because of an AI chatbot , and 71% say AI meaningfully affected which products or brands they seriously considered or chose at least half the times they used it to shop . This confirms AI recommendations are actively reshaping brand switching behavior, not just assisting research.

Are consumers comfortable letting AI agents complete purchases automatically?

Comfort levels remain cautious. Research found 53% are uncomfortable allowing an AI assistant to make purchases on their behalf , while only 7% would permit autonomous purchasing under predefined conditions . Most shoppers want AI to inform decisions through recommendations and price comparisons rather than fully automate the final transaction.

The New Algorithmic Gatekeepers Demand a New Playbook

The toy aisle data point is small in scope but massive in implication. It proves that AI assistants are not neutral search engines returning objective results, they are opinionated gatekeepers running on rules most marketing teams have never tested. Matt Britton has built his reputation as a AI keynote speaker precisely because he translates findings like this into clear, board-ready action plans rather than abstract warnings.

Brands that wait for a standardized measurement framework to emerge will likely wait too long. The companies moving now, testing their visibility across Alexa for Shopping, ChatGPT, Google AI Mode, and Copilot, are the ones setting the benchmarks competitors will chase for the next several years. Matt Britton's keynote platform and his book Generation AI give executive teams a practical starting point for that audit.

For leaders who want a deeper, ongoing read on these shifts, Britton unpacks new data every week on The Speed of Culture podcast, and brands can pressure-test their own AI visibility using the consumer intelligence tools inside Suzy. The retailers winning share of model today are quietly building the advantage that will define market share for the rest of the decade. Book Matt Britton to help your leadership team see the rules before your competitors do.