What changed
46% of US AI-recommended buyers returned an item; 36% said it differed from the description
When
Survey fielded September 2026, released October 27, 2026
Who it affects
Brands selling on Amazon, Walmart and their own sites as shoppers use AI assistants
What to do
Audit listing attributes, size charts and images for anything an AI could misread

Key takeaways

  • 46% of US shoppers who bought on an AI recommendation returned an item in the past year, per Blue Yonder's September 2026 survey.
  • The top reason, 36%, was that the product differed from its description, which points to missing or vague listing data.
  • Fill every attribute, put measurements in text, test Rufus answers and keep holiday return terms simple.

Share of US shoppers who bought on an AI recommendation

  • Gen Z43%
  • Millennials36%
  • Gen X19%
  • Baby boomers9%
Source: Blue Yonder 2026 Consumer Retail Returns Survey, Oct 2026

What the survey found

Blue Yonder, a supply chain software company that also sells returns management tools, polled more than 5,000 consumers in the US, UK, France, Germany and Australia, including 2,000 Americans, through a third-party provider in September 2026. In the US, 24% had bought a product after an AI recommendation. Of those, 46% had returned an item in the past year, compared with 41% for purchases driven by social media recommendations. Among people who returned an AI-recommended item, 36% said it was different than described and 23% said the size or fit was wrong. Nearly half of them, 49%, said they are more likely to return AI-recommended items than items they found themselves.

The surprise: it is a description problem, not a fit problem

Across all US returns, size and fit is the top reason at 42%, and 'different than described' is only 20%. For AI-driven purchases the order flips. Blue Yonder CEO Duncan Angove put it plainly: "It's a data problem." His point, reported by Retail Customer Experience, is that assistants recommend from whatever product information they can read, and most product copy was written for a search bar and a human skimming images, not for a machine matching a shopper's request. When an attribute is missing, vague or wrong, the AI fills the gap and the shopper finds out at the door. Only 20% of US respondents trust AI to pick the right size.

Why AI-recommended items came back

36%different than described
23%wrong size or fit
20%trust AI on size
63%refuse a return fee
Source: Blue Yonder 2026 Consumer Retail Returns Survey, Oct 2026

Who is buying this way

Younger shoppers lead. 43% of Gen Z and 36% of millennials have bought on an AI recommendation, against 19% of Gen X and 9% of baby boomers, and 56% of Gen Z AI shoppers returned an item. These are the same shoppers most sensitive to return rules: 47% of Gen Z stopped shopping at a retailer that tightened its policy, and 29% of all US respondents did. Retail Dive also cited a SmartCustomer survey finding 89% of shoppers do not fully trust AI recommendations, so the trust gap is wide even as usage grows. On Amazon this matters through Rufus, which answers from your listing, reviews and Q&A; see our Rufus optimization guide.

So what for a brand's money

A return costs more than the refund. On Amazon you lose the referral fee portion Amazon keeps, pay return processing on many categories, and often get back an item you cannot resell as new. Returns also feed your return rate and listing health signals, as explained in our guide to Amazon return reasons. The good news in this survey is that the biggest cause, a mismatch between the listing and the product, is fixable by the brand. Tightening policies is the wrong lever: 51% of US shoppers picked a retailer for its returns policy, and 63% refuse to pay a return fee.

Worked example with illustrative numbers

These numbers are invented to show the arithmetic. A brand sells 3,000 units a month of a $40 product. If 10% of orders now come from AI assistant recommendations, that is 300 orders. At an assumed 15% return rate on those orders, 45 come back; at 9%, the brand's rate on other orders, it would be 27. The 18 extra returns each cost about $14 in lost fees, return processing and unsellable stock, roughly $252 a month, or about $3,000 a year on one ASIN. Rewriting attributes and the size chart is a one-time job that takes a few hours.

What sellers should do this week

  • Listing manager: in Seller Central, pull the Voice of the Customer and return reports for your top 20 ASINs and tag every 'not as described' comment. Metric: count of not-as-described returns per 100 orders.
  • Listing manager: fill every structured attribute in the listing, including material, dimensions, capacity, compatibility and count, so an assistant does not have to guess from the title. Our Amazon listing optimization service starts here. Metric: attribute completeness in the listing quality dashboard.
  • Content lead: put exact measurements and a size chart in text, not only in images, and state what the product does not do. Metric: return rate on the ASIN 30 days after the change.
  • Brand manager: ask Rufus and one outside assistant five real shopper questions about your product and note every wrong or vague answer. Metric: share of correct answers, rechecked monthly.
  • Customer service lead: answer the questions that keep coming up in reviews and Q&A in the bullets, because assistants quote them. Metric: repeat questions per week.
  • Owner: keep return terms simple through the holidays rather than tightening them. Metric: conversion rate and return rate side by side through January.

Make your listing AI-proof

  • Every attribute filled
  • Measurements in text
  • Size chart in text
  • Say what it does not do
  • Answer top review questions
  • Test Rufus monthly

When this does not apply

The survey counts shoppers who returned at least one AI-recommended item, not the share of AI-driven orders returned, so it does not tell you your own AI return rate. Blue Yonder sells returns software, which gives it a reason to highlight returns. Brands in categories where fit is not an issue, such as consumables, will see far fewer description returns. And if your returns are mainly damage or late delivery, fix packaging and shipping first.

Where PPC Spider helps

We rewrite listings so both shoppers and AI assistants read the same facts, and track the return rate as part of every account review. Follow the ecommerce marketplace news channel, or start with a free Amazon audit.

Sources

Get the next change before your competitors do

We send the marketplace changes that matter each week, with what to do next.

By subscribing you agree to our privacy policy. Unsubscribe any time.
Want this done for your account?

We run these playbooks on client accounts every day. Start with a free audit.

Get a free audit