AI Shopping Returns Surge: Inside Retail's New Trust Crisis

Nearly half of American shoppers who bought something because an AI assistant told them to have already sent it back. That single statistic, pulled from Blue Yonder's newly released 2026 Consumer Retail Returns Survey, exposes the most urgent problem in retail right now. AI shopping returns are climbing faster than most Fortune 500 insights teams can explain them, and the reason has nothing to do with the intelligence of the technology itself.

Nearly half of U.S. shoppers, 46%, who bought an item based on an AI recommendation have returned one item in the past year. That number should stop every retail executive mid-sentence. Brands spent 2025 racing to deploy chatbots, agentic shopping assistants, and generative search integrations. Few of them paused to ask whether the product data feeding those systems was actually good enough to earn a customer's trust.

Matt Britton has spent the better part of two decades studying how consumer behavior shifts the moment new technology enters the shopping journey. As founder of Suzy, a real-time consumer intelligence platform, and author of Generation AI, Britton has tracked the gap between what brands deploy and what consumers actually experience. He argues this returns data is the clearest early warning sign yet that agentic commerce is scaling without the infrastructure to support it.

This is not a story about AI failing. It is a story about brands outrunning their own data hygiene, and customers paying the price in wasted time, return shipping, and eroding trust. Matt Britton frequently tells Fortune 500 audiences that the winners in this next commerce cycle will not be the companies that adopt AI fastest. They will be the ones that fix the data underneath it first.

What the Blue Yonder Data Reveals About AI Shopping Returns

The Blue Yonder survey is the most rigorous look yet at how AI recommendations translate into real purchase behavior. Now in its fourth year in the U.S., the survey polled more than 5,000 consumers across the U.S., Australia, France, Germany and the U.K., including 2,000 Americans. The findings are not a fringe anomaly. They represent a structural pattern emerging across every major retail market.

The top reasons for shoppers returning an AI-driven purchase are that it was different than described and/or an incorrect size of fit. In other words, the AI told shoppers something that was not true, or it recommended a size without the data to back up the suggestion. Blue Yonder CEO Duncan Angove was blunt about where the blame belongs. "Shoppers are taking AI's advice faster than retailers are getting ready for it," Angove said , adding that the real issue sits upstream of the algorithm itself.

The fit problem is especially revealing. Just one in five U.S. consumers trusts AI to recommend the right size and fit, and more than eight in 10 would cut back at a retailer that limits their returns. That combination is dangerous for retailers. Consumers do not trust the recommendation, yet they still buy, and when the product arrives wrong, they punish the retailer rather than the AI tool that misled them.

Matt Britton points out that this dynamic inverts the usual innovation adoption curve. Normally skepticism slows purchase behavior until trust builds. Here, purchase behavior is outpacing trust entirely, which means the damage shows up after the sale rather than before it.

The Agentic Commerce Trust Gap: Why Data Quality Is the Real Problem

Industry analysts have started using the phrase "agentic commerce trust gap" to describe this exact dynamic, and the root cause keeps tracing back to the same source. "An AI assistant is only as good as the product information it can find, and most product data was written for a search bar, not for a machine suggesting products for someone," Angove said. That distinction matters enormously. A product title optimized for a search engine is not the same as a structured, machine-readable data set that an AI agent can reason over accurately.

Independent testing backs this up. Core dna's test of 20 popular Black Friday products found that ChatGPT got about half of its answers wrong, quoting prices stores weren't charging, naming retailers that didn't stock the product, or pointing to a different item altogether. A separate audit found the problem is even more widespread across platforms. A September study from Product.ai put the same 220 shopping questions to free and paid versions of ChatGPT, Claude, Gemini and Perplexity, and found reproducible factual conflicts on prices, model names and specs in 86% of questions.

Consumer confidence reflects this instability. Another survey found that 89% don't fully trust AI recommendations, even though most have used AI to make a purchase in the past year. That means shoppers are proceeding with purchases despite deep skepticism, which sets up an almost inevitable return. Rithum's research across US and UK shoppers found that more than half blame the retailer or brand when an AI recommendation contains incorrect product information.

This is the piece Matt Britton emphasizes most on stage during his AI keynote presentations: consumers do not separate the retailer from the AI tool in their minds. If the bot gets it wrong, the brand absorbs the reputational cost, not the technology vendor. That shifts the responsibility for data accuracy squarely onto the retailer, whether the C-suite has budgeted for it or not.

Generational Divide: Gen Z Leads AI Adoption and AI Returns

The data gap is not distributed evenly across age groups, and that matters for any brand targeting younger consumers. Gen Z (43%) and millennials (36%) respondents are more likely to have purchased a product after receiving a recommendation from an AI tool or assistant, compared with only 19% of Gen X and 9% of baby boomers. Younger shoppers are the early adopters of agentic commerce by a wide margin, which means they are also absorbing most of the friction.

That friction shows up directly in return rates. Returns are highest among the youngest shoppers, with more than half of Gen Z consumers who bought on an AI suggestion have sent at least one of those items back in the past year. Ironically, Gen Z is also the generation most willing to extend trust to the technology in theory. While 20% of overall respondents said they would trust an AI tool to recommend the right size or fit, that trust ranges from 36% of Gen Z respondents to only 9% of baby boomers.

This creates a paradox Matt Britton has written about extensively in Generation AI: the generation most optimistic about AI's potential is also the generation most frequently burned by its current execution gaps. Brands courting Gen Z loyalty through AI-powered shopping experiences need to recognize that a single bad recommendation can undo months of brand equity. This is precisely the kind of generational nuance Britton unpacks for audiences during his Gen Z keynote speaker engagements, where he connects behavioral data to practical retention strategy.

Loyalty behavior confirms the stakes. Gen Z (47%) and Millennials (40%) were more likely to have stopped shopping at a retailer that implemented stricter returns policies in the past year, compared with Gen X (24%) and Baby Boomers (17%). Younger consumers will not tolerate a retailer that recommends the wrong product and then punishes them for returning it.

The Hidden Cost: Return Policies, Loyalty, and Margin Risk

The instinct for many operations leaders will be to tighten return policies as volume climbs. Blue Yonder's data suggests that response would backfire badly. Flag shoppers as frequent returners and limit their returns, and eight in 10 say they'd cut back or walk away. Angove described the core tension plainly: "The instinct when returns go up is to tighten the rules. But the data tells you how that ends."

The challenge is distinguishing abuse from loyalty. The difficulty is telling the difference between a shopper who abuses returns and a loyal customer who simply returns a lot. A heavy returner driven by AI-caused sizing errors looks identical in the data to a serial abuser unless the retailer tracks the root cause of each return.

Return policy also directly shapes acquisition and retention economics. Over half of U.S. respondents (51%) said that they have decided to buy from one retailer over another purely because it had a better returns policy. That means returns policy has become a competitive differentiator rather than a back-office cost center, and AI-driven mis-recommendations are now a direct input into that calculation.

Margin exposure compounds quickly here. Processing a single return costs far more than the shipping label suggests once restocking, repackaging, and markdown risk are included. Matt Britton frequently reminds finance and retail leaders during his finance industry keynotes that AI-driven revenue gains on the front end mean little if they are offset by rising reverse-logistics costs on the back end.

How Fortune 500 Retailers Can Close the AI Trust Gap

Closing this gap requires treating product data as core infrastructure rather than a metadata afterthought. Matt Britton outlines several priorities for insights and retail leaders working to get ahead of this problem:

• Audit product data for machine readability, not just search engine optimization, since most catalog data was never built for an AI agent to reason over.
• Build sizing data pipelines that connect return reason codes directly back to product listings so AI tools learn from actual fit failures.
• Separate frequent returners from AI-caused returners in CRM systems before applying punitive return policies.
• Test AI assistants against real purchase scenarios quarterly, the way Product.ai's 86% error-rate study did, rather than trusting vendor claims.
• Train insights teams on AI consumer behavior in 2026 so return-rate anomalies get traced to root cause instead of blamed on customer fickleness.

Real estate, fashion, and finance brands are each facing a version of this same data quality problem, just with different product complexity. Britton addresses industry-specific nuances like these during his real estate industry keynotes, where property descriptions and AI-generated listings carry similarly high stakes for buyer trust. The underlying lesson transfers across every vertical: AI recommendations are only as trustworthy as the data feeding them.

Retailers serious about closing this gap should also study how consumer sentiment evolves in real time rather than relying on quarterly retrospectives. Platforms like Suzy allow brands to test product data, sizing language, and AI-generated descriptions against live consumer panels before mistakes compound into returns. Matt Britton built this capability specifically because waiting for a Blue Yonder-style survey to surface a trust crisis is already too late for the brands living inside it.

Key Takeaways for Business Leaders

• Audit product catalog data for AI readability before expanding any agentic commerce deployment further.
• Separate return reason codes by channel so AI-driven returns get tracked distinctly from traditional returns.
• Resist the urge to tighten return policies broadly, since punitive measures risk losing loyal customers over AI-caused errors.
• Invest in real-time consumer testing of AI recommendations before scaling them across the full catalog.
• Prioritize Gen Z and millennial shopper experience, since this cohort drives both the highest AI adoption and the highest return volume.

Frequently Asked Questions

Why are AI shopping returns increasing in 2026?

Nearly half of U.S. shoppers who bought an item based on an AI recommendation have returned it, mainly because the product was different than described or the wrong size. The core issue is data quality, not AI intelligence. Product catalogs were built for search engines, not for AI agents making specific recommendations to individual shoppers.

What is the agentic commerce trust gap?

The agentic commerce trust gap describes the disconnect between how fast retailers deploy AI shopping assistants and how little consumer trust those tools have actually earned. Nearly 89% of consumers don't fully trust AI recommendations, even though most have used AI to make a purchase in the past year. That gap shows up as returns, complaints, and lost loyalty.

Which consumers are most affected by AI product recommendation errors?

Gen Z shoppers are hit hardest, with more than half who bought on an AI suggestion sending at least one item back in the past year. Gen Z and millennials also adopt AI shopping tools at far higher rates than older generations, making them the most exposed to data quality failures.

How can retailers reduce AI-driven product returns?

Retailers should prioritize machine-readable product data, connect return reason data directly back to catalog listings, and test AI-generated recommendations against real consumer panels before full deployment. Punishing frequent returners without separating AI-caused errors from abuse risks losing loyal customers entirely, since eight in 10 shoppers say they would cut back after a retailer limits their returns .

The window for fixing this quietly is closing fast. Every quarter that passes with AI shopping returns climbing unchecked is a quarter of eroded trust that becomes harder to win back. Matt Britton has built his reputation helping Fortune 500 leaders see these inflection points before they become headlines, translating consumer data into decisions that protect both margin and brand loyalty.

The brands that treat this moment as a data infrastructure problem, not a technology problem, will be the ones still earning consumer trust when agentic commerce matures. Those that keep scaling AI without scaling the data behind it will keep absorbing the returns, the costs, and the churn. For organizations ready to get ahead of this shift, Matt Britton's speaker platform and The Speed of Culture podcast offer a direct line into where consumer trust in AI is heading next.