AI recommends your product. Only 5% buy on the spot. The rest go check your reviews.

THE 5% PROBLEM

According to a Feedvisor survey reported by Modern Retail and Digiday on 23 October 2026, ChatGPT is where 16% of shoppers start looking for products, but only 9% go there to actually buy. When an AI assistant recommends a product, only 5% say they buy it on the spot. The rest check other platforms, like Google and Amazon, for better prices first.

So the recommendation gets you on the shortlist. It rarely closes the sale.

THE NEXT FOUR TABS

PartnerCentric asked shoppers what they do after an AI suggestion:

• 33% check existing reviews.
• 22.3% compare prices.
• 18.8% search Amazon or another retailer.
• 14.8% visit the brand's website.
• 7.6% look for creator content about the product.

And for holiday buying, 33% of shoppers say shipping and return policies often or always change what they buy.

Read that list from the shopper's side. Reviews first. Price second. Then another store. Your own site comes fourth.

EVEN THE BELIEVERS CHECK

You might think shoppers who trust AI just buy what it says. They don't. In PartnerCentric's data, 42.2% of shoppers in the two highest AI-trust groups still want reviews before acting, and 31.2% check reviews right away.

"AI is a starting point, not a final verdict," said Stephanie Harris, founder and CEO of PartnerCentric. "So the recommendation has to survive the next question the shopper asks. And that question usually gets answered in the reviews, on the retailer page, or in the shipping and return terms."

People also limit what they let AI do for them. A Prosper Insights & Analytics survey cited in the same report found 12.1% would let an AI agent book a restaurant, 11.0% would let it buy groceries, and only 7.3% would let it buy apparel.

WHY THIS IS A CONVERSION PROBLEM

CRO, conversion rate optimisation, means getting more of the people who arrive to take the action you want. Usually you have to guess what stops them. Here the surveys tell you: reviews, price, other retailers, shipping and returns.

A worked example with round numbers: say AI assistants send you 1,000 interested shoppers in a month. At 5%, 50 buy on the spot. The other 950 go checking. If your reviews, prices and returns win back just 10% of them, that is 95 more orders. Total: 145, nearly triple.

THE FIVE-STEP PAGE FIX

1. Reviews next to the price. Show the star rating and review count near the buy button, not at the bottom.
2. Fresh reviews. Ask every buyer for a review about a week after delivery. Old reviews make a product look old.
3. Price proof. If you are cheaper than Amazon, say so. If you aren't, say what they get by buying direct.
4. Shipping and returns above the fold. Cost, delivery date and return window in one plain line.
5. Creator proof. Put short creator or customer videos on the product page, where the 7.6% go looking anyway.

THE PUNKD TAKE

Everyone is chasing AI mentions like they're the finish line. The mention just starts the fight. The sale goes to whoever wins the next three tabs the shopper opens: reviews, price, shipping. Fix the page they land on, then go get cited.

STEAL THIS

1. Search your product name plus "reviews" today and read the first page like a stranger would.
2. Move rating, review count, delivery date and returns next to the buy button.
3. Set up an automatic review request a week after delivery.
4. Compare your price and story on Amazon with your own site, and close the gaps.
5. Add two short creator videos to your best-selling product page.

When an AI sends someone to check you out, what do they find?