By Steve Merrill, Founder of WRKNG Digital | June 23, 2026
When AI shopping agents can't recommend your products, it's almost always a data problem. not a traffic problem, not a budget problem. We audited over 2,400 Shopify products across dozens of stores. Only 11% had feeds clean enough to be recommended by ChatGPT Shopping, Google AI Overviews, or Perplexity. Here are the 10 mistakes that disqualified the other 89%.
1. Missing GTINs
GTINs (Global Trade Item Numbers) are how AI agents and shopping platforms cross-reference your product against a known catalog. Without one, Google Merchant Center flags your listing as incomplete, and ChatGPT Shopping skips it entirely when matching user queries to products. Fix: Add GTINs to every product variant in Shopify's barcode field. If you manufacture your own products, register for a GS1 prefix at GS1 US and generate valid GTINs.
2. Vague Product Titles
Titles like "Blue Shirt" or "Men's Tee. Summer" don't match how AI agents parse intent. They need structured, attribute-rich titles: Brand + Product Type + Key Attribute + Size/Color. A title like "Patagonia Men's Organic Cotton Crew Neck Tee. Navy Blue, Large" gives an AI agent five distinct matching vectors. Vague titles get ignored when the agent is comparing options for a specific query.
3. Non-Normalized Variant Attributes
If one variant says "Small," another says "S," and a third says "sm," your feed looks inconsistent to automated systems. Google's product data spec expects standardized attribute values. and AI agents trained on that spec behave the same way. Pick one format per attribute and enforce it across your entire catalog. Non-normalized attributes are one of the top reasons feeds get partially rejected by shopping channels.
4. Outdated Pricing
AI agents pull price data to answer "what's the cheapest option for X." If your feed price doesn't match your storefront price, you'll get flagged for price mismatch. and removed from active recommendations until the discrepancy clears. This happens constantly with sale pricing that expires. Fix your Shopify feed sync frequency. If you run time-limited promotions, use the sale_price_effective_date field so the agent knows when the deal ends.
5. Low-Image-Quality Flags
AI shopping channels use image quality signals to filter out products that won't convert. Google's guidelines require images to be at least 100x100px for non-apparel (250x250px for apparel), but the real threshold for AI recommendation eligibility is closer to 800x800px. No watermarks, no placeholder images, no lifestyle-only shots without a clean product image. A product with a low-resolution hero image gets deprioritized automatically.
6. Missing Category Taxonomy
Shopify doesn't require you to map products to Google's product taxonomy. AI agents do. The Google Product Taxonomy has over 6,500 categories. If your product isn't mapped to one, agents can't place it in the right context when a user asks for recommendations in a specific category. Map every product type. It's tedious once. It pays off every time an agent responds to a category-specific query.
7. No Product Descriptions
AI agents read descriptions to extract features, materials, use cases, and compatibility. An empty or one-line description means the agent has almost nothing to work with when a user asks "what's a good [product] for [specific use case]." Write 150–300 words per product. Include materials, dimensions, who it's for, and what makes it different. That content is what gets surfaced in an AI-generated comparison or recommendation.
8. Inconsistent SKUs
SKUs are how your products stay tied together across systems. your feed, your fulfillment, your storefront. When SKUs change between platforms or variants share a SKU by mistake, the feed breaks. AI shopping channels that cache product data treat a changed SKU as a new product. That resets any reputation or review signals attached to the old one. Lock your SKU structure early and don't change it.
9. Absent Inventory Signals
If your feed doesn't include availability status (in_stock, out_of_stock, preorder), AI agents default to assuming availability is unknown. and they won't recommend an unknown. ChatGPT Shopping and Perplexity both filter out products without clear availability signals when generating purchase recommendations. This field takes five minutes to configure in your feed. It's not optional.
10. Mismatched Brand Names
The brand name in your feed has to match the brand name on your product pages, in your schema markup, and in your Google Merchant Center account. When they don't match, AI agents can't reliably verify the source. and verification matters when they're recommending a product to a user who asked for a specific brand. Even minor inconsistencies ("WRKNG Digital" vs. "Wrkng Digital" vs. "WRKNGDigital") cause mismatches. Pick one canonical brand name and use it everywhere.
How This List Was Built
These 10 mistakes came from direct feed audits across Shopify stores we've analyzed using WRKNG Digital's AI readiness scoring system, cross-referenced against Google Merchant Center's feed diagnostics documentation and observed disqualification patterns across ChatGPT Shopping, Google AI Overviews, and Perplexity. These aren't hypothetical. they're the actual reasons stores we've audited were invisible to AI agents before fixing them.
FAQ
Q: Do I need to fix all 10 before my products show up in AI recommendations?
No. Start with GTINs, title structure, and availability signals. those three unlock the most channels fastest. Work through the rest systematically. You'll see improvement as each issue clears, not just at the end.
Q: Which AI shopping channels are most affected by product feed quality?
Google AI Overviews, ChatGPT Shopping (via Bing product index), and Perplexity Shopping all pull from structured feed data. Each has slightly different thresholds, but the core requirements. GTINs, clean titles, taxonomy, availability. are shared across all three.
Q: Can Shopify fix these issues automatically?
Shopify's native feed is a starting point, not a finished product. Apps like Feedonomics or DataFeedWatch can automate normalization, taxonomy mapping, and title reformatting. but someone still needs to configure the rules. The data problems in your catalog don't fix themselves.
Q: How long does it take AI channels to recognize feed improvements?
Google re-crawls approved feeds every 24–48 hours. ChatGPT Shopping typically refreshes within 72 hours of a feed update. Most stores see measurable AI visibility improvements within a week of fixing the top three to five issues.
Q: What's the easiest way to audit my current feed for these issues?
Start with Google Merchant Center's diagnostics tab. it flags most of these directly. For AI-specific gaps (category taxonomy, description depth, GTIN coverage), you need a separate audit that evaluates your feed against AI recommendation criteria, not just Google's ad requirements.
If you want to know exactly where your store stands on AI readiness, get the WRKNG Digital AI Commerce readiness assessment. We'll show you which of these mistakes your feed has right now. and what fixing them is worth.
