By Steve Merrill, Founder of WRKNG Digital | July 2, 2026
We audited 2,400 Shopify products across a mix of stores earlier this year. Only 11% had enough structured data to qualify for recommendation by an AI shopping assistant like ChatGPT Shopping or Google AI Overviews. That's internal data from WRKNG Digital, 2026.
That means 89% of those products are functionally invisible to the AI tools now driving product discovery.
The problem isn't your SEO. It isn't your ad spend or your content calendar. It's your product feed. Most merchants have no idea it's happening.
What AI Shopping Tools Are Actually Reading
AI shopping assistants don't browse your Shopify storefront. They read structured product data.
When someone asks ChatGPT "What's the best waterproof hiking boot under $150?", the AI pulls from product feeds. The same catalog data you submit to Google Merchant Center and Microsoft Shopping. It cross-references that data against the query, checks product identifiers against known databases, and decides whether your product is a confident enough match to recommend.
If the data is missing, thin, or inconsistent, the AI doesn't surface your product. It just moves on.
It's not a penalty. There's no algorithm update to recover from. The AI simply doesn't have enough information to recommend your product with confidence, so it doesn't.
The Audit That Changed How We Think About This
When we ran our 2,400-product audit at WRKNG Digital, we expected maybe 40 or 50 percent of products to have gaps. The reality was far worse.
67% were missing a GTIN or MPN entirely. 78% had product descriptions under 100 words. 54% had no structured product_type taxonomy. Nearly a third had title formats so vague that an AI couldn't determine the product category without inferring it from the image.
These aren't edge cases. These are the defaults you get when you build a product catalog for human shoppers browsing a website. That format doesn't translate to AI.
The merchants running these stores weren't doing anything wrong. They just never had a reason to think about feed quality before now. That reason exists.
The Fields That Determine AI Recommendation Eligibility
Not every feed field carries equal weight. Here's what actually moves the needle.
GTIN and MPN are product identity fields. A GTIN (Global Trade Item Number) or MPN (Manufacturer Part Number) lets AI systems verify your product against known product databases and cross-reference it with reviews, specs, and comparison data. Google Merchant Center's product identifier requirements treat these as required for most product categories. Without them, your product is anonymous. No cross-referencing happens. No confidence score gets built.
Product title structure is where most merchants silently lose. "Blue Mountain Fleece" tells an AI almost nothing. "Patagonia Better Sweater Fleece Jacket Men's Medium Navy" tells it the brand, product type, key feature, demographic, size, and color. That's six matchable attributes in one field. The format that works every time: Brand + Product Type + Key Attribute + Variant Descriptor.
Description length and density matter more than most people realize. AI models use your product description to match queries. A 30-word description produces almost no signal. A 150-plus word description with material composition, dimensions, use case, compatibility details, and care instructions gives the AI enough context to match your product to a specific query. In our audit, products that met AI recommendation thresholds averaged 210 words in their descriptions.
The product_type taxonomy field is the one most merchants either leave blank or fill in inconsistently. "Apparel" is useless. "Women's Outerwear > Waterproof Jackets" is useful. AI shopping tools use product_type to filter and categorize before they evaluate your specific product. Flat or vague taxonomy means you're filtered out early in the matching process.
The brand field seems obvious, but it's often wrong. It needs to match exactly how the brand appears in external databases. Capitalization and spacing matter. If you're a private label brand, consistency across your entire catalog matters more than the specific name.
Availability and price consistency are about trust signals. AI tools cross-reference your feed data against your actual storefront. If your feed says "in stock" but your product page shows "sold out," that inconsistency flags your feed as unreliable. Google Merchant Center actively disapproves products with mismatched data between feed and landing page. Stale or inconsistent data erodes confidence the same way missing data does.
Fix These Fields First
You don't need to fix everything at once. If you're starting from zero, here's the order that produces the fastest results.
First: GTINs and MPNs for your top 20% of products by revenue. These are your highest-intent items. Getting them properly identified unlocks cross-referencing and comparison eligibility immediately. Shopify stores your GTIN in the "Barcode" field under each product variant — that's what flows into your Google Merchant Center feed via Shopify's Google channel integration.
Second: Title rewrites for your top 50 products. Use the Brand + Product Type + Key Attribute + Variant format. This is a spreadsheet job, not a development project.
Third: Description expansion for those same 50 products. Push to 150-plus words. Add specifics — what it's made of, who it's for, how it's used, what it fits or works with. Specificity is the signal.
Fourth: product_type taxonomy cleanup. Pick a consistent taxonomy structure and apply it across your catalog. Google's product taxonomy is a useful starting point — it's what most shopping feeds default to.
Start with your top revenue products. Don't try to fix 2,400 listings before you see what fixing 50 does.
How to Fix Your Shopify Product Feed for AI Visibility
These are the three steps we walk every merchant through in our audits.
Step 1: Run a feed completeness audit. Export your product catalog from Shopify (Products > Export) and open it in a spreadsheet. Check four columns: barcode/GTIN, description word count, product_type, and title structure. Flag every product missing any of the four. That's your work list. Sort by revenue descending.
Step 2: Rewrite product titles for AI parsing. Take your flagged top 50 products and rewrite titles using the structured format. Measure before and after using impressions in Google Merchant Center as a proxy for feed pickup. You'll see movement within 1 to 2 weeks.
Step 3: Expand descriptions to 150-plus words with specific attributes. Pull your shortest descriptions and add what matters for matching: material, dimensions, use case, compatibility, care instructions. You're writing for a system that needs enough signal to confidently match your product to a query. Every added attribute increases that confidence score.
Frequently Asked Questions
Does fixing my product feed affect AI shopping tools beyond Google?
Yes. Google Merchant Center feeds into Google's AI Overviews and Shopping features. ChatGPT Shopping and Microsoft Copilot Shopping pull from Bing's product index, which uses similar feed quality signals. Cleaner feed data improves your position across all of them, not just Google.
My products don't have GTINs because we're a private label brand. Is there a workaround?
Use MPNs that you define yourself and set the identifier_exists field to "no" in your Google Merchant Center feed. This tells the feed processor you're a brand without universal product codes. It doesn't eliminate the disadvantage entirely, but it avoids feed disapprovals and keeps your products eligible for AI recommendation rather than getting filtered out.
How long should a product description be for AI visibility?
Our audit found that products meeting AI recommendation thresholds averaged 210 words. The floor is 150 words, but content quality matters more than hitting a number. A 150-word description full of marketing copy ("You'll love this amazing jacket!") performs worse than a 100-word description packed with material specs, dimensions, and use cases. Attributes beat adjectives.
How often should I update my product feed?
Google Merchant Center requires updates at least every 30 days to keep products active, but recommends daily updates for price and availability data. AI shopping tools favor feeds with current, consistent data. Set up automated feed refresh through Shopify's Google channel app or a third-party feed tool. Stale data signals low reliability, same as missing data.
Is this the same as regular Shopify SEO?
No. Traditional Shopify SEO focuses on storefront pages: title tags, meta descriptions, URL structure, on-page content. Product feed quality targets shopping-specific indexes and is an entirely separate system. You can have excellent on-page SEO and a broken product feed. They don't fix each other.
The Window Is Open Right Now
The brands that will dominate AI shopping recommendations over the next two years aren't the ones with the biggest ad budgets. They're the ones with the cleanest, most complete product data.
89% of the products we've audited aren't there yet. That's a real opening for brands that move now, before AI shopping becomes as competitive as paid search.
If you want to see exactly where your Shopify store stands — including your feed quality score, which fields are missing, and a prioritized fix list — the WRKNG Digital AI Commerce Readiness Assessment covers it in full. We built it for Shopify merchants who want to know the truth about their AI exposure before the window closes.
