6 Product Feed Fields AI Shopping Assistants Check Before Recommending Your Store

By Steve Merrill, Founder of WRKNG Digital | July 4, 2026

AI shopping assistants don't browse your store. They read your product feed — and six specific fields decide whether your product gets recommended or skipped entirely.

Most Shopify stores are sending feeds that were built for Google Shopping in 2019. That was fine then. It's a problem now. ChatGPT Shopping, Microsoft Copilot, Perplexity Shopping, and Google's AI Overviews all evaluate feeds differently than a traditional search crawler does. They're pattern-matching against user intent. And if your data doesn't give them a clear match, they move to a competitor who does.

1. Product Title (With Full Attributes)

Your product title isn't just a name. It's the primary signal an AI uses to match your product against a user's query. "Blue Hoodie" tells an AI nothing useful. "Men's Midweight Fleece Hoodie — Navy, 100% Recycled Polyester, S–3XL, Cold-Weather Hiking" tells it exactly what it needs to match against "warm eco-friendly hoodie for hiking."

According to Google's Merchant Center product title guidance, titles should front-load the most descriptive attributes: material, color, size, and use case. AI shopping layers sit on top of these same feeds. What Google requires, AI assistants depend on.

Good looks like: brand + product type + key differentiator + variant attributes. Bad looks like whatever Shopify auto-generated from your variant title three years ago.

2. Product Description (150+ Words, Benefit-Led)

AI assistants pull from product descriptions when they need to answer follow-up questions. "Is this machine washable?" "Does it work for wide feet?" "What's the weight capacity?" If that information isn't in your description, the AI either guesses or drops your product from consideration.

We ran 2,400 products through our AI Commerce Readiness audit tool. Only 11% had descriptions long enough and detailed enough to support AI recommendation. The rest were under 50 words, feature-led instead of benefit-led, and contained none of the specifics an AI would need to answer a real customer question.

The bar is 150 words minimum. Lead with what the product does for the customer. Follow with specs. End with fit, care, and compatibility details. That structure gives AI something to work with.

3. Price and Currency (Accurate to the Cent)

AI shopping assistants are being used for purchase decisions — not just discovery. That means price accuracy isn't optional. ChatGPT Shopping and similar tools surface price prominently in responses. If your feed price is stale, out of sync with your storefront, or listed as a range without a base value, the AI either flags the inconsistency or drops your listing.

"Contact for price" is an automatic disqualification. AI assistants won't recommend products they can't price. That's a trust and liability issue on their end — and your invisible problem on yours.

Price should match your storefront to the cent. Currency must be explicitly declared (USD, CAD, EUR — not implied). Sale prices need start and end dates in your feed, or the AI may treat them as permanent and create a mismatch when users land on your site.

4. Availability and Inventory Status (Real-Time)

This one kills recommendations silently. An AI assistant recommending an out-of-stock product gets punished — users click, hit a dead end, and lose trust in the platform. So platforms have gotten aggressive about filtering out stale availability data.

Google's product availability field specification requires one of four values: in_stock, out_of_stock, preorder, or backorder. Most Shopify stores feed this accurately for the main variant — but break down on variant-level inventory. If your black size-M is in stock but your navy size-L isn't, and your feed doesn't reflect that at the variant level, AI platforms see ambiguous data and drop the product.

Feed refresh frequency matters here too. A feed that updates every 24 hours can't keep up with flash sales, sellouts, or restock events. Aim for 6-hour or real-time feed updates if your inventory moves fast.

5. Product Category (Mapped to Google Shopping Taxonomy)

AI shopping assistants categorize products to route queries correctly. "Show me running shoes under $120" only works if your running shoes are actually filed under running shoes — not "Footwear > Athletic > General" or some custom taxonomy you invented.

Google's Shopping taxonomy contains over 6,000 categories. It's the de facto standard most AI shopping platforms are built on top of. Using it correctly — at the most specific level possible — is what allows AI to match your products to category-filtered queries.

A store selling "Women's Trail Running Shoes" should map to Apparel & Accessories > Shoes > Athletic Shoes > Running Shoes, not just Apparel & Accessories > Shoes. The more specific the mapping, the more query types your product becomes eligible for.

6. GTIN / Barcode (The #1 Trust Signal for AI Product Matching)

GTIN — Global Trade Item Number — is the barcode on your product. For AI shopping assistants, it's the single most important trust signal for product identity verification. When a GTIN is present, an AI can cross-reference your product against manufacturer data, pricing databases, and review aggregators. It knows your product is real, identifiable, and consistent across sources.

Without a GTIN, your product is an island. The AI can't verify it matches what you say it is. According to Google Merchant Center's GTIN requirements, products with GTINs consistently receive better placement and are eligible for features that GTIN-less products are excluded from — including AI-powered shopping surfaces.

If you manufacture your own products, you need to register for GTINs through GS1. If you resell branded products, the GTIN is on the packaging. Either way, leaving this field blank is leaving your products off the AI's verified list.

How We Know This

We built an AI Commerce Readiness audit tool and ran it across 2,400 Shopify products from 38 stores. We scored each product against the field-level requirements used by ChatGPT Shopping, Copilot, Perplexity, and Google AI Overviews. Only 11% of products cleared all six fields. The most common failures: short descriptions (74% of products), missing GTINs (61%), and titles with no attribute data beyond product name (58%).

The stores scoring highest on AI readiness also had the cleanest Google Shopping performance — which confirmed what we suspected. AI shopping is built on top of the same data infrastructure as traditional product search. The difference is the tolerance for bad data dropped to near zero.

Frequently Asked Questions

Do all AI shopping assistants use the same product feed fields?

Not exactly the same, but they draw from the same standards. Google's Merchant Center spec is the dominant reference point. ChatGPT Shopping, Copilot, and Perplexity all index feeds or structured data that aligns with that specification. Getting these six fields right covers you across platforms.

My Shopify store already has a Google Shopping feed. Does that mean I'm covered?

Only if your feed is current and accurate at the field level. Many Shopify stores generate feeds automatically, but auto-generated titles strip attributes, descriptions stay at factory defaults, and GTIN fields get left blank. A feed existing isn't the same as a feed working for AI.

How often should I update my product feed?

For stable inventory, daily is fine. For stores with fast-moving stock, flash sales, or frequent price changes, 6-hour refreshes prevent AI platforms from surfacing stale data. Most Shopify feed apps support custom refresh intervals — set it and forget it.

What happens if AI shopping assistants find an error in my feed?

They either demote the product or exclude it entirely. There's no penalty box with a warning. The platform's job is to give accurate results to users. Bad feed data makes your product unreliable to surface. You don't get notified — you just disappear from recommendations.

Is structured data on my product pages a substitute for a clean product feed?

No. They're complementary. Structured data (Product schema on your PDPs) helps AI crawlers understand your pages. Product feeds give AI shopping platforms a structured, queryable dataset they can match against user intent in real time. You need both.

Want to see how your product feed scores? Get your free AI Commerce Readiness audit at WRKNG Digital.