You ask ChatGPT, "What's the best insulated water bottle for the gym?" It names three brands. None of them are yours — even though you sell exactly that, and you rank fine on Google.

This is the frustrating part of AI shopping: getting your catalog connected to ChatGPT is automatic, but getting recommended is not. Shopify's Agentic Storefronts syndicate your products to AI channels by default. The plumbing works. Whether the AI actually picks you is a different question entirely.

Here are the five reasons your store gets skipped, the fix for each — and, just as importantly, how to tell whether your fixes are working, because that last part is where most guides go quiet.

First, understand how ChatGPT chooses

Google returns ten links and lets the shopper pick. ChatGPT returns an answer — usually one to three products. There's no page two. If you're not in that short list, you don't exist for that shopper.

That makes AI recommendation "winner-take-most." The stores that win are the ones the AI can read, understand, and trust. Almost everything below comes back to one of those three.

Reason 1: Your product data is too thin

This is the most common failure, and the most basic. If your product page says "high-quality hoodie" and little else, ChatGPT has nothing to work with — no fabric, no fit, no use case, no buyer type. It can't confidently recommend a product it doesn't understand.

The fix: Write product descriptions in natural language that answer the questions a shopper would actually ask — material, sizing, use cases, care, who it's for. Think about the phrasing people use when they talk to an assistant ("best X for Y"), not just keywords. Add an FAQ to your product and collection pages. Our full guide on how to write Shopify product descriptions AI assistants recommend walks through the exact structure.

Reason 2: You're missing structured data (schema)

AI agents rely on structured data to parse your product's attributes, price, and availability with confidence. Yet by most estimates only around 18% of e-commerce product pages have complete schema markup. Without it, the AI has to guess — and it usually guesses another store.

The fix: Add JSON-LD structured data to your theme — Product schema on product pages, Organization schema site-wide, and FAQPage schema where relevant. You can check any page with Google's Rich Results Test.

Reason 3: Your GTIN / barcode is empty

This one is overlooked constantly. Without a GTIN (barcode), AI shopping systems have to match your product by name and description alone, which is unreliable and often just fails.

The fix: Fill in the barcode field on your products (Settings → Products in Shopify), and make sure your JSON-LD includes it in the correct GTIN format.

Reason 4: Your robots.txt is blocking AI crawlers

Some Shopify themes and apps block the crawlers that AI platforms use to verify your product data. Your store can be connected through Shopify's feed and still be unreadable, because the AI can't visit your actual pages to confirm details.

The fix: Check your storefront's robots.txt and make sure crawlers like GPTBot, ClaudeBot, PerplexityBot, and Google-Extended aren't disallowed. A plain-text llms.txt file summarizing your business also helps AI assistants understand what you sell — though for Shopify stores in 2026 there's a twist worth knowing about first.

Reason 5: You have no trust signals

AI systems learn from the wider web. If your brand has almost no reviews, creator mentions, Reddit discussions, or press, the AI has little reason to trust — and therefore recommend — you.

The fix: Build review volume, get products in front of creators, and earn mentions in niche publications. This is the slowest fix, but it compounds.

The part everyone skips: how do you know it worked?

Here's the problem with every "5 reasons" checklist (including, honestly, the one you just read): you can do all five fixes and still have no idea whether they moved the needle. AI recommendations aren't a dashboard you can refresh. So most merchants "optimize," wait, and hope.

You need two feedback loops:

1. Visibility — are you actually being recommended now? The manual version: ask ChatGPT, Claude, Gemini, and Perplexity to recommend a product in your niche, and note whether you appear and who does instead. Do this before and after your fixes. It's tedious to do by hand across four engines and multiple products, but it's the only way to know if "read and understand" improved.

2. Revenue — are the recommendations turning into sales? Visibility without sales is a vanity metric. The real test is whether AI-referred shoppers are buying — and, as we've written about separately, Shopify's built-in attribution misses a large share of those orders. You want to see actual order revenue tied back to each AI channel, not just a session count.

This is exactly the gap Vanto fills. It checks, product by product, whether ChatGPT, Claude, Gemini, and Perplexity recommend your store — giving you a clear before/after on visibility — and it reads your real Shopify orders to show which AI channels are actually driving revenue. It also flags the fixable stuff (weak descriptions it can rewrite for you, whether AI crawlers can reach your store, an llms.txt generator) so the "fix" and the "did it work" live in one place.

The bottom line

Getting listed in ChatGPT is automatic. Getting recommended comes down to whether the AI can read your data, understand your products, and trust your brand — and the five fixes above cover the first two almost entirely.

But don't stop at fixing. The merchants who win this channel are the ones who measure — checking whether they're being recommended, and whether those recommendations turn into real orders. Optimizing blind is just hoping with extra steps.

Curious whether AI assistants recommend your products right now? Vanto runs the check across ChatGPT, Claude, Gemini, and Perplexity — and ties it back to real Shopify revenue. Free to start, right inside Shopify.