By Steve Merrill, Founder of WRKNG Digital | June 30, 2026
Only 11% of Shopify Products Can Be Recommended by AI
We ran 2,400 Shopify products through our AI visibility audit. Only 11% had the data structure needed to be recommended by ChatGPT Shopping.
Not 11% of bad stores. Not 11% of small stores. Products pulled from established Shopify merchants - stores with real traffic, real reviews, real revenue.
The other 89% are invisible. Not because their products are bad. Because their data is wrong.
What the Product Data Gap Actually Means
There's a gap between what merchants put in their product listings and what AI assistants need to make a recommendation. That gap is what's killing your AI visibility.
Google's crawler could figure out your product from context. It would read your page, infer your product type, and piece together the attributes from your description copy. AI assistants don't work that way. ChatGPT Shopping, Perplexity, and Microsoft Copilot pull from structured data signals. If the signal isn't there, the product isn't there.
I've seen stores with beautiful products, strong reviews, and healthy traffic from Google - completely invisible to ChatGPT and Perplexity because their product data is a mess. The products exist. The data doesn't.
The 6 Fields AI Assistants Actually Check
When an AI shopping assistant evaluates whether to recommend your product, it's looking at a short list of structured signals. These six fields determine whether you make the cut.
1. Product type - specific, not vague. "Apparel" doesn't help. "Men's Crew Neck T-Shirt" does. AI assistants match against category taxonomies when generating recommendations. Vague types either get dropped or get misclassified.
2. Material, size, and color attributes - explicit values, not buried in descriptions. "Available in multiple sizes" is useless. "Sizes S, M, L, XL, 2XL, 3XL" is a signal. AI assistants filter by attribute. If your attributes aren't populated as discrete fields, they don't count.
3. Real-time availability. AI shopping assistants increasingly pull live inventory data. A product marked "in stock" with no availability feed attached gets deprioritized. According to Google's Merchant Center documentation, availability signals are weighted heavily in AI-powered shopping surfaces.
4. Price with currency code. Not just a number. "29.99" without a currency code creates ambiguity AI systems resolve by skipping the product. Price must be formatted with ISO 4217 currency codes - USD, CAD, GBP - attached to the value.
5. Brand name - exact, not your store name. Your store is called "Blue Ridge Goods." Your brand is "Patagonia" or "your own brand name." These are different fields. AI assistants matching brand queries against product feeds need the brand field populated with the actual brand, not the merchant name.
6. GTIN or UPC where available. For products that have them, the Global Trade Item Number is a direct match key. ChatGPT Shopping and Google Shopping Graph use GTINs to deduplicate and verify product identity. Missing GTINs don't disqualify a product - but present GTINs significantly boost confidence scoring.
Why Most Stores Are Getting This Wrong
Shopify's default product setup doesn't force you to think in AI-readable fields. You write a title, write a description, maybe pick a product type, and move on. That workflow was fine for SEO-era commerce. It's not fine for AI-era commerce.
The most common failure pattern in our audit data: generic titles, empty product type taxonomy, and blank attribute fields.
"Blue T-Shirt" is a product title. "Men's Pima Cotton Crew Neck T-Shirt, Navy Blue, Sizes S-3XL" is a data record. AI assistants work with data records. They can't infer what you didn't say.
The second failure pattern: merchants fill in the Shopify product type field with internal category names that mean nothing outside their own store. "Summer Collection" is not a product type. "Women's Swimwear" is. Shopify supports Google's product taxonomy - 6,000+ standardized categories. Most merchants never touch it.
The Compound Effect of Missing Fields
Based on our internal audit data, each missing structured field reduces AI recommendation probability by approximately 15-20%. That math compounds fast.
A product missing a specific product type, two explicit attributes, and a brand field isn't 60% less visible. The signals interact. An AI assistant can't confidently match a vague product against a specific user query when multiple confirmation signals are absent. The product gets skipped for one with cleaner data, even if that competing product is objectively worse.
Perplexity's shopping results - documented in their 2025 shopping rollout - explicitly weight structured data completeness as a ranking factor. ChatGPT Shopping, which pulls from Bing's shopping index, follows similar logic.
Clean data beats good products in AI-powered discovery. Right now, that's the reality.
How to Fix It - Prioritized by Impact
Fix these in order. The top of this list has the most use per hour of work.
First: Rewrite product titles to be explicit. Include product category, primary material or differentiator, key variant attribute (color or size range), and gender/age group if applicable. This single change moves the needle more than anything else we've tested.
Second: Map product types to Google's taxonomy. Shopify lets you set a Google product category field directly in the product admin or through your Google channel settings. For your top 20% of products by revenue, set the most specific matching category you can find. "Apparel & Accessories > Clothing > Shirts & Tops > T-Shirts" beats "Clothing" every time.
Third: Populate attribute metafields. Shopify's metafield system lets you add structured attributes - material, fabric weight, fit type, care instructions - that feed directly into your product data export. These aren't visible to customers unless you surface them. But they're visible to AI systems reading your feed.
Fourth: Add brand to every product. If you're selling your own brand, create a brand metafield and populate it with your brand name - not your store name - consistently across all products. Consistency matters. "WRK" and "WRKNG" are different strings to a machine.
Fifth: Add GTINs where you have them. For branded products with UPCs, enter them. Shopify has a dedicated barcode field. Use it. This is especially important if you're selling established brands alongside your own products.
According to Google Merchant Center's required product data guidelines, these fields are table stakes for inclusion in AI-powered shopping surfaces - which Google, Microsoft, and OpenAI all feed from in some form.
This Is Not an SEO Problem
I want to be direct about something. The stores in our audit with the worst AI visibility scores often had the best traditional SEO. Well-written descriptions. Good keyword density. Clean URL structures.
None of that helps you get recommended by an AI shopping assistant. SEO optimizes for crawl and rank. AI shopping optimizes for structured data and attribute matching. They're different problems with different solutions.
The stores winning in AI-powered discovery right now have invested in their data infrastructure - not their content. That's a shift most merchants haven't made yet. Shopify's own AI commerce research points to structured product data as the primary driver of AI shopping performance.
The window to get ahead of this is open. It won't stay open long.
Frequently Asked Questions
How do I know if my store has the product data gap?
Run your top 50 products through a structured data check. Look for missing product type taxonomy, blank attribute fields, and product titles that don't include category or material. If more than half have gaps, your store has the problem.
Does this affect all AI shopping assistants or just ChatGPT?
All of them. ChatGPT Shopping (via Bing), Perplexity Shopping, Google AI Overviews shopping, and Microsoft Copilot commerce features all pull from structured data signals. The specific fields weighted may vary slightly, but the core six fields matter across all platforms.
Do I need to rebuild my entire product catalog?
No. Start with your top 20% by revenue. Those products drive the majority of your sales and search volume. Getting them right has an outsized impact. Then work down the catalog over time using Shopify's bulk edit tools or a feed management app.
How long does it take to see results after fixing product data?
Feed refresh cycles vary by platform. Google Shopping index typically refreshes within a few days of a feed update. Bing's index (used by ChatGPT Shopping) can take 1-2 weeks. Perplexity's product data sources update on rolling cycles. Expect 2-4 weeks to see meaningful improvement after a data overhaul.
What Shopify apps help with product data quality?
Shopify's native Google channel handles basic feed submission. For attribute management at scale, apps like Simprosys Google Shopping Feed or DataFeedWatch give you more control over field mapping. But the data quality problem is a content problem first - no app can write accurate product attributes for you.
Find out if your store has the product data gap → Get a free audit
