.avif)
Why AI Shoppers Buy Long-Tail Products
Three-quarters…75%. That is the share of AI-attributed orders on Shopify in the second quarter that came from outside the company’s top 100 product categories, according to Shopify’s president, Harley Finkelstein, speaking to analysts on the company’s earnings call on August 5. He called it the “sweet spot”. He also said AI-driven traffic and orders to Shopify stores had tripled from a year earlier, and that half of AI-referred sessions land directly on a product page — two and a half times the rate for traditional search.
If you run a specialty store, read that again. The orders AI is sending are not going to the hundred biggest categories. They are going to everything else — the “long tail” of the catalog, which is where most of us live.
What The Long Tail Is And Why AI Shoppers Land There
Picture every product sold online lined up by how often it sells. At the left end sit the few hundred things everyone buys — the phone cases, the running shoes, the standing desks. Stretching off to the right, for miles, sit the millions of things that each sell a little: the desk sized for a 400-square-foot apartment, the car seat that lets three ride across the back of a sedan. That long, thin stretch is the long tail. Any one product in it sells rarely. Added together, the tail outsells the head — and that is where a specialty store makes its living.
Search engines were built for the head. Type “standing desk” and the result is a ranked list of the biggest sellers, because that is what most people typing those two words want. An AI assistant is built for the tail. Nobody asks it for “shoes.” They ask for the desk that fits the small apartment, or the car seat that fits three across — Finkelstein’s own examples — and the assistant is designed to answer a specific question with a specific product. A specific question almost never lands in a top-100 category. It lands on a page that happens to describe exactly the thing the shopper asked for.
That is why half of those sessions arrive on a product page rather than a homepage. The assistant is not sending a browser to your store. It is sending a buyer to one product.
I spent part of my twenties as a buyer for a toy chain, and the top categories ran the building. The big names — the Cabbage Patch Dolls, Nintendo 8-bit, whatever owned the end-caps that year got the planogram meetings, the co-op dollars and the arguments. Everything else in the catalog — the doll cradles, the replacement power supplies and controllers (e.g. the specific, odd, hard-to-find thing a parent drove across town for) — got a shelf tag and no attention, because no one at headquarters was searching for it. The parent was. That is the long tail, and it was there forty years before anyone put an AI in front of it. What changed is that the parent now asks a machine, and the machine finds the shelf tag if the shelf tag says enough.
What The Number Does Not Say
Honest limits, because the figure travels better than its footnotes.
Shopify did not publish the list, the ranking rule behind “top 100,” or the baseline — what share of all its orders fall outside those categories. Without the baseline, 75% cannot be read as AI over-indexing on niche products; it may simply mirror the shape of Shopify’s catalog. It is one quarter, from one company that benefits from the story. And it says nothing about Google: Shopify itself described AI search as augmenting Google, not replacing it.
What it does say is enough: AI buyers arrive on product pages, in volume, for specific products. Which makes the product page the thing to get right.
The Product Page Is Where This Is Won Or Lost
Adobe — which owns the Magento commerce platform — in an April 2026 analysis of more than a trillion visits to U.S. retail sites, scored page types on how readable they are to large language models. Homepages scored 75 out of 100, category pages 74, and product pages 66 — the weakest page on the site is the one AI shoppers land on. Adobe’s authors put it plainly: much of that product-page content is currently invisible to the models, because it sits in images, tabs and scripts the models do not read.
Google is signalling the same thing from the other side. Its Merchant Center added an attribute completeness score in May that shows which products lack the structured attributes — color, material, size, style — shoppers ask for in AI Mode and AI Overviews. Google does not say how heavily those attributes weigh, only that it is now measuring them for you.
Recomaze, a vendor of AI-visibility services, ran 9,720 stores through Gemini on six purchase-intent questions each and found the stores that won every query were mostly focused, single-category brands, and that AI “rewards categories where products have clear, describable specifications.” One engine, one scan, a vendor’s study — but it points the same way as Shopify and Adobe: specific, describable products on readable pages.
How Do I Know If AI Shoppers Can Find My Products?
Ask an assistant the way a buyer would, and see whether you come up. Five minutes, no tools.
Write down three questions a real customer would ask about your products — not your category, your products. “Which dishcloth holds up in the dishwasher,” not “kitchen textiles.”
Ask each question to two AI models (ChatGPT and Gemini are enough). Note which stores and products are named.
For any product of yours that was not named, open its page and read only the plain text — not the photos, not the tabs. Could a stranger answer the three questions from that text alone?
If not, the fix is not a tool. It is the missing sentence: the size, the material, the use, the thing that makes it the answer to a specific question.
We ran this on October 10, logged out of ChatGPT so nothing it knew about us could help.
Asked which Swedish dishcloth holds up in the dishwasher and where to buy one, it named two small specialty stores alongside Amazon — one of them a store with a few hundred products and no advertising budget, whose product pages happen to say, in plain text, what the cloth is made of and how to wash it.
Asked which shooting rest to buy for a short range bench, it named the manufacturer and two large retailers and no small store at all, even though several sell the same bags; their pages describe the product in a sentence, and the manufacturer’s page gives the dimensions, weight and the non-slip base. Same question type, opposite outcome, and the difference was on the page.
A product page that answers a specific question in plain text is the whole game here. The 75% is not about being big. It is about being specific. Long-tail products win when their pages are specific.
The Simple Path to Better Visibility
Pick your ten best-selling long-tail products and run the three-question test on each. Where a page fails, add the sentence that answers the question — in the description, as text, where a model can read it. That is an hour of work, it costs nothing, and it is the same work that makes a product easier for a human to buy. Then set a reminder for the Q3 call: Shopify will report the number again, and you will know whether your store was in the 75%.
Jim Boudreau has worked in SEO & eCommerce for twenty-eight years.
