AI Recommendations Raise the Stakes on Product Detail Pages
Shoppers are turning to AI assistants such as ChatGPT, Perplexity and Amazon’s Rufus for product recommendations. Those assistants draw on product detail page (PDP) content, so the quality of that content reaches the shopper before the purchase. Returns have many drivers, such as fit, price, impulse buying, etc. PDP content is one that marketers can influence directly.
Blue Yonder’s 2026 Consumer Retail Returns Survey polled 2,000 Americans in September. Of the 24% of US consumers who have bought on an AI recommendation, 46% returned an item in the past year.
Among US shoppers who returned an AI-recommended purchase, 36% said the item differed from its description, the most common reason.
Search results sent shoppers to the product page, where they read the details. An AI assistant condenses the same listing into one short answer. Any gap in the listing travels into that answer, and the shopper builds an expectation from it. Materials, dimensions and fit notes carry more weight when the answer may be the main thing a shopper reads.
When returns climb, the instinct may be to shorten return windows or add stricter limits. Doing so can cost goodwill.
A University of Tennessee study tested this with 1,500 US consumers across three experiments. A restocking fee or a shorter return window made participants less willing to buy from the retailer or recommend it. Even shoppers who rarely return items often reacted negatively. With policy changes costing shoppers’ goodwill, the more useful fix sits earlier, in what the product page promises.
A first step is auditing the most-returned items against what an AI answer would say about them, covering materials, dimensions and fit notes. Return reasons logged at the point of return show which listings to check first. Take a jacket listed as water-resistant. If the listing never says how much rain it handles, an AI answer may repeat the phrase as written. A shopper buying for a rainy commute may then send it back.
Whoever writes a listing is now writing for answer engines too, which puts it on the marketing team’s list. Marketing teams can give product copy the same scrutiny they give an ad, since an AI answer may repeat it as written.
Each listing can become the script for an AI answer, and that answer may be the first description a shopper reads.
Leave a Reply
You must be logged in to post a comment.