Short answer: Across 25 AI-readiness reports on independent specialty stores, three findings kept repeating. The assistants know these stores by name but recommend big competitors in anonymous category searches; the products show up in ChatGPT shopping while the sale routes to resellers, often because barcodes are missing from the feed; and the in-store expertise that makes each store special is invisible to machines. All three have free fixes.
Over the past few weeks we've been running our AI-readiness assessment on independent specialty stores — ski and backcountry shops, curl-care brands, natural-fiber apparel makers, running stores, instrument dealers. For each one, we read the site the way an AI buyer agent would, asked Claude, ChatGPT, and Gemini about the store live (with web search, the way a shopper's assistant actually works), and ran a live test of ChatGPT's shopping surface.
Twenty-five stores is a small sample, and specialty retail is a particular corner of commerce. But three findings showed up so consistently — across verticals, platforms, and store sizes — that they're worth sharing. None of this is theory; every one of these came out of a real store's report, usually more than once.
Why does AI know my store but recommend competitors instead?
Being known and being recommended are two different things, and the gap between them is where independent stores are quietly losing the next generation of shoppers. Here's the pattern that surprised us most. When we asked the assistants about a store by name — "what can you tell me about [store]?" — they almost always knew it. Often in flattering detail: the founding story, what the store is known for, even its reputation for fitting or formulation. These stores are not invisible.
Then we asked the same assistants the question that actually matters: the category question, with no store name. "I'm looking for backcountry ski gear — where should I buy?" "What should I use for my curly hair?"
In almost every case, the store that the assistant had just described so warmly was nowhere in the answer. The recommendations went to the big names: national retailers, celebrity brands, marketplaces.
Why does this happen? Assistants recommending "the best X" lean heavily on what the rest of the web corroborates — best-of lists, review sites, high-authority mentions. Big brands dominate that corpus. A beloved local shop with a 25-year reputation can have almost no machine-readable evidence of it.
The free fix: make the evidence exist. Structured data that links your business to its social profiles and reviews, an about page that says plainly what you're known for, FAQ pages that answer the questions your staff answer daily. None of this guarantees a recommendation — but without it, the assistant has nothing to cite even when it wants to.
Why does the sale go to resellers when my products are in ChatGPT shopping?
Your products can be fully eligible and present in ChatGPT shopping while the sale still goes to someone else, because the products are there but the merchant isn't. Several stores we assessed had done everything their platform asks: catalog feed working, every required policy published, fully eligible for ChatGPT's shopping surface. Some even showed up in our live tests with their products and direct links — genuinely ahead of the pack.
And yet, when we asked ChatGPT to shop those stores' categories, a pattern kept appearing: one brand's products showed up sold by resellers on two other continents. A running store's inventory appeared — credited to other shops. Eligible, present, and not getting the sale.
One mechanical reason turned out to be measurable, and it floored us. We checked the product feeds of three stores — a founder-formulated beauty brand and two running retailers — for barcodes (GTIN/UPC), the identifier that lets a listing attach to the canonical product card where shopping AIs compare sellers and hand out the sale. The coverage was 0 of 87 variants, 0 of 380, and 0 of 504. Zero, in all three. Including retailers whose products have the UPC printed on the physical box — the data simply never made it into the feed.
Without identifiers, your listing can still be found when a shopper names your store. But in the anonymous searches where most demand starts, it floats unattached — while the product card it should be on lists your competitors.
The free fix: the barcode field in your platform admin is sitting empty. Fill it. It won't guarantee ranking — nobody outside these companies can promise that — but it removes a hard blocker your competitors may not even know exists.
Why can't AI see what makes my store special?
The thing that makes you special — a real consultation almost every store we assessed has already built — is exactly what no AI can see. This one is less a defect than a heartbreak. That consultation is the thing AI shopping needs most, and it takes real forms:
- Treadmill gait analysis and 3D foot scans
- Boot fitting refined over decades
- Curl-typing guidance by porosity and pattern
- Individual neck measurements on every guitar in the building
- Human chat, staffed by people who genuinely know the gear
And essentially none of it is visible to a machine. The fit process happens in-store and vanishes online. The chat widget says "we'll be back tomorrow" to a shopping agent that needed an answer in four seconds. One shop even runs an AI chatbot on its own site — a real investment, genuinely useful to visitors — and it's just as unreachable to an outside buyer agent as the treadmill is, because it only exists for people who already found the website.
So when a shopper asks their assistant "which shoe fits an overpronator with wide feet?", the answer comes from a manufacturer's marketing copy — not from the store that answers that exact question thirty times a day, better than anyone.
The free fix, and its honest limit: write the knowledge down where machines can read it. Turn the questions your staff answer daily into real FAQ and guide pages with structured data. That gets your expertise quoted. What it can't do is the back-and-forth — the "tell me about your hair" or "what's your boot sole length?" that makes a consultation a consultation. Static pages answer questions; they don't ask them. That conversational layer — an agent of your own that knows your products and can actually talk to a shopper's AI — is the missing piece across every store we assessed. It barely exists in the wild yet. It's the gap we started sqrly to fill, and honestly, it's why we ran these assessments in the first place.
What should I do if I sell online?
- Check what the assistants say about you — by name, and then by category without your name. The second answer is your real competitive position.
- Open your product feed and look for barcodes. If the field is empty, you have an afternoon of free work with real upside.
- Inventory your invisible expertise: every question your team answers in person, on the phone, or in chat is content — and eventually, conversation — that no competitor can copy.
The stores we assessed aren't behind because they did something wrong. They're behind because the web quietly grew a second kind of visitor — one that reads structured data instead of vibes, and asks questions instead of browsing. The good news in all three findings is the same: the raw material is already there. It just isn't speaking the new visitor's language yet.
Want to see all three findings for your own store? Run your free AI-readiness report — it's the same assessment we used here, and your email is the whole price.
sqrly is building the seller-side agent that closes that gap — a salesperson for your store that knows your products the way your best floor staff do, for every shopper and every AI agent. We're pre-launch, working with a small group of founding merchants. If that's a conversation you want to be part of, join the waitlist.