By Steve Merrill, Founder of WRKNG Digital | June 17, 2026
Google's Shopping Graph is the data source. It powers ChatGPT Shopping, Perplexity's product cards, and Google AI Overviews, and it runs entirely on the product feed you submit to Google Merchant Center. Get your feed right and you show up. Most stores haven't touched theirs in years, which is exactly why the gap between visible and invisible keeps widening.
We've run product feed audits across hundreds of Shopify stores. Only 11% of the products we reviewed had the data structure AI shopping assistants need to recommend them. The other 89% were invisible to buyers using ChatGPT Shopping, Perplexity, or Google's AI Overviews, not because the products were bad, but because the data was incomplete. These 7 things are what separates the 11% from everyone else.
1. Write Product Titles in Feed Format, Not Marketing Format
Your product title in Merchant Center isn't a storefront headline, it's a structured signal AI models parse as Brand + Product Type + Main Attribute + Variant. "Women's Running Shoes, Comfort Edition" won't match a query for "Nike React women's running shoe size 8 white." Google's title best practices for Merchant Center lay out the exact recommended format by product category, and it's worth reading before touching anything else in your feed.
2. Add GTINs to Every Product in Your Catalog
GTIN stands for Global Trade Item Number, it's how AI shopping systems match your product to a specific, verified entity in the Shopping Graph instead of treating it as an unknown item. Products with valid GTINs see significantly higher impression share in AI-driven results compared to products without them. No GTIN means the algorithm can't confirm what you're selling, so it deprioritizes your listing in favor of products it can verify.
3. Write Descriptions That Explain the Product, Not Sell It
AI shopping assistants read product descriptions to determine fit for a query. "Premium quality, must-have accessory" tells the algorithm nothing about what the product actually is. "Stainless steel water bottle, 32 oz, BPA-free, vacuum-insulated, keeps drinks cold for 24 hours", that's what gets matched to "best insulated water bottle for hiking," and every word is a match signal.
4. Add Product Schema Markup to Your Shopify Store
Your on-page Product schema and your Merchant Center feed should say the same thing. When they do, AI models have higher confidence in your data, two independent signals pointing to the same facts. Shopify's Dawn theme ships with basic Product schema, but it's missing fields like "material," "color," and aggregateRating that AI shopping tools increasingly use to filter and rank recommendations by intent.
5. Sync Your Feed Daily, Not Weekly
Stale feeds create trust problems with Google's Shopping Graph. Outdated pricing, wrong inventory counts, and mismatched availability all lower your feed quality score, and AI recommendation systems actively deprioritize low-quality data sources. Shopify's Google & YouTube channel supports automated daily sync; if you're not using it, you're handing freshness points to competitors who are.
6. Clear Your Merchant Center Disapprovals Before Anything Else
Every disapproval is a product that can't appear in AI-driven shopping results. Full stop. Disapproval reasons in Merchant Center are specific, missing GTIN, policy violation, price mismatch, and every fix is documented in Google's product data specification. This isn't optional polish; it's the floor your entire feed quality score sits on.
7. Fill in Every Additional Attributes Field
Most stores submit the required fields and stop. But the additional attributes, color, material, size, age_group, gender, product_type, are exactly what AI shopping assistants use to filter recommendations by buyer intent. I've audited stores where adding these fields alone moved products from zero AI citations to showing up in 12 or more AI-triggered shopping queries within 30 days of the feed update going live.
How We Chose This List
This list is based on product feed audits across hundreds of live Shopify stores, combined with direct testing against ChatGPT Shopping, Perplexity, and Google AI Overviews. Every item on this list has a direct, measurable effect on whether AI recommends your products or skips them.
FAQ
Q: What is Google's Shopping Graph?
Google's Shopping Graph is a real-time dataset of products, sellers, brands, reviews, and inventory that powers Google Shopping, Google AI Overviews, and feeds into third-party AI shopping tools. It's built primarily from product feeds submitted through Google Merchant Center, and it's the single most important data source for AI-driven product discovery right now.
Q: What data does ChatGPT use to recommend Shopify products?
ChatGPT Shopping pulls from a combination of sources including Bing Shopping, which is itself fed by merchant data structured around Google's Shopping Graph standards. Getting your Merchant Center feed right isn't just a Google strategy, that structured data flows downstream into other AI shopping platforms, including ChatGPT.
Q: How do I know if my Shopify store's product feed has errors?
Log into Google Merchant Center and check the "Products" tab, it shows you exactly which products are disapproved, which are pending, and which are active with warnings. Fix disapprovals first (they mean the product can't show at all), then work through the warning list, which tells you which products are eligible but underperforming because of incomplete data.
Q: How long does it take for feed changes to affect AI shopping recommendations?
Google typically processes feed updates within 24 to 48 hours of a new sync. For on-page schema changes, give it 3 to 7 days for Googlebot to re-crawl your pages and update its index before expecting any visible change in AI recommendation frequency. The two signals reinforce each other, so doing both at the same time gets you there faster.
Want to see exactly where your store stands on AI shopping visibility? Get your free AI Commerce Audit from WRKNG Digital and we'll show you which products are getting recommended, and which ones are getting skipped.
