AI got basic facts wrong for 64% of retailers. What is it telling shoppers about your store?

Shoppers now ask ChatGPT, Gemini and Perplexity about stores before they buy. A new study checked how often those answers are right, and the results should make any store owner look at what AI says about them.

What the study found

Searchable asked ChatGPT, Gemini and Perplexity more than 72,000 questions about UK high street retailers. Each answer was checked against the retailer's real details. Here is what they found:

• 64% of the retailers had at least one false fact in an answer about them.
• The assistants answered 98% of the questions, and about one answer in 16 was wrong.
• The most common mistake was a wrong postcode, in about one location answer in 10.
• About one website answer in 15 sent shoppers to a dead link, a lookalike site or a different business.
• Perplexity got 10% of its answers wrong, Gemini 5% and ChatGPT 4%.

The assistants almost never said "I don't know." They answered nearly every question, right or wrong, in the same confident tone.

Smaller stores were hit hardest

Searchable found that smaller retailers suffered most. Its co-founder, Chris Donnelly, said AI mistakes are more likely to hit smaller businesses than larger brands. The reason he gave: a big brand is written about in many places online, so AI has plenty of correct information to draw on. A small shop may have little more than its own website and a Google listing.

If your store is mostly known through its own site, that site is most of what AI has to go on.

What the study didn't check

The questions were about locations, contact details and brand names. They didn't ask about products, prices or return policies. So the study doesn't tell us how often AI gets your prices or your returns window wrong.

It does show something useful. These were simple, public facts, like an address, and AI still got them wrong for most retailers. Product details change more often. Prices go on sale, items sell out, and policies get updated.

There are real examples. A UK marketing consultant, Lilach Bullock, heard from a prospect who had asked ChatGPT about her. ChatGPT said she charged from £3,000 a day. She never had. She traced the number to an old directory listing. After the listing was corrected, she says the wrong price stopped showing up in AI answers within about three weeks.

Why this matters to shoppers

A separate survey asked shoppers about wrong AI answers. Rithum surveyed 1,046 online shoppers in the US and UK:

• 58% said they blame the retailer or brand when an AI recommendation has wrong product information.
• 16% said they would avoid buying the product entirely after a bad recommendation.
• 67% said price is the most important thing AI has to get right.
• Only 5% said they go straight to the retailer's or brand's website to check what AI told them.

So when AI gets your store wrong, most shoppers don't check with you. And many of them hold you responsible.

Check what AI says about your store

You can see this for yourself in a few minutes. Ask ChatGPT, Gemini and Perplexity the questions a shopper would ask before buying from you:

• "What is [your store]'s return policy?"
• "How much does [your best-selling product] cost at [your store]?"
• "Does [your store] ship to [a place you ship to]? How long does it take?"
• "What is [your product] made of, and what sizes does it come in?"

Then compare each answer with what your store says. Note where the answer came from if the assistant shows its sources. A wrong answer often comes from an old page, a marketplace listing or a review site, not from your store.

Where wrong answers start

AI can only repeat what it finds. If your product page leaves out the material, an assistant may fill the gap from somewhere else. If your shipping policy lives only in your Shopify settings, an assistant reading your product page may not see it. We explained how those two places can disagree in Two kinds of structured data, and why the studies seem to disagree.

The fix starts with your own product data: complete product details, current prices and stock, and your shipping and return policies written where AI looks for them. When your store is the clearest and most complete source, AI has less reason to guess.

To see what an AI assistant can read on your product pages, try our free check.

What this doesn't settle

Both studies come from companies that sell services in this area, and neither has published its full method. The Searchable study covered UK high street retailers, not online stores, and didn't test product facts. And no store controls everything AI says: assistants also read reviews, forums and other websites. What you can control is whether your own store gives AI the right answer to find.