By Steve Merrill, Founder of WRKNG Digital — July 20, 2026

What product data fields decide whether AI recommends your products?

AI recommends products it can read, match, and trust, and that comes down to nine data fields: title, description, unique identifier, brand, category, price, availability, images, and attributes like size and material. Fill these well and you're eligible. Leave them blank and you're invisible.

We ran a catalog audit on a client store last month. Most products were missing three of these nine. That's the gap between showing up and staying hidden.

1. Product title

The title is the first thing AI reads. Write it the way a buyer searches: product type, key attribute, color, and who it's for. "Cast iron 12-inch skillet, pre-seasoned" beats "The Chef's Choice."

2. Product description

The description carries the details AI uses to compare. Materials, dimensions, use case, care. Put real specifics in plain sentences. Google's product structured data guide shows how these details map to what search and AI systems read.

3. Unique identifier (GTIN or MPN)

This is the field that matters most. A GTIN or MPN lets AI match your product to the exact item a shopper named. Without it, you're a guess. With it, you're a match.

4. Brand

Brand tells AI who makes the item and whether it fits a branded query. "Best Yeti-style tumbler" versus "Yeti tumbler" are different questions. A clear brand field puts you in the right one.

5. Product category

Category places your product in the right shelf. Use the standard taxonomy Shopify and Google share so AI knows a "running shoe" is footwear, not apparel. Miscategorized products get filtered out.

6. Price

Price has to be present and accurate in your Offer schema. Shoppers ask AI for products in a range. No price means you can't answer "under $100," so you drop out of that query.

7. Availability

Availability keeps you from recommending something that's sold out. AI checks stock status before it suggests a product. Keep it live and accurate through your feed.

8. Images

Clean images with descriptive alt text help. AI systems increasingly read visual data, and alt text gives the model a text handle on what the image shows. Empty alt tags waste the signal.

9. Attributes (size, color, material)

Attributes answer the picky questions. Size, color, material, fit. These are what a shopper narrows on, so they're what AI narrows on. Fill the variant data across every option.

How do you fix all nine at scale?

Here's the hard part. Fixing nine fields on one product is easy. Doing it across 2,000 products is a grind. That's where most stores stall out. Validate against the Schema.org Product spec, then work the catalog in batches by category.

I made this mistake myself early on. I fixed the top sellers and figured that was enough. It wasn't. AI reads the whole catalog, not your bestsellers.

FAQ

Which field matters most?

A unique identifier like GTIN or MPN. It lets AI match your product to the exact item a shopper asked for and compare it fairly.

Does the description really affect recommendations?

Yes. AI reads it for materials, use case, and fit. Write it for a human buyer and you give AI what it needs.

Do I need all nine filled?

No, but more is better. Missing a required identifier or price can disqualify a product outright.

How do I check my current data?

Run a feed audit that scores completeness by field across the full catalog. Sampling hides the gaps.

Want your catalog scored?

We check all nine fields across your entire Shopify catalog and show you exactly what's missing. Get your product data audit here.