By Steve Merrill, Founder of WRKNG Digital | September 14, 2026
AI assistants read your product title, description, price, and availability first. Everything else is a tiebreaker. If those four fields are weak, the model moves on before it ever sees your material or your reviews.
I ran an ecommerce business to $10M a year, and our feed looked complete on every old checklist. It still said almost nothing a machine could use to recommend us. The fix wasn't more fields. It was better ones, in the order that matters.
What does an AI assistant actually read from a Shopify feed?
It reads the fields that answer a shopper's question. When someone asks an assistant for "a warm merino base layer under $120," the model scans titles for the product type, descriptions for the material and use, price for the budget, and availability to make sure it can actually recommend something in stock.
It does not read your whole catalog line by line. It pattern-matches against the question and grabs the products whose top fields line up. Miss on the first four, and you're out of the running.
Title comes first, and most stores waste it
The product title is the single most-read field in your feed. It's the first thing the model uses to decide if you match the query at all.
"Merino Wool Base Layer, Men's, 200gsm" gets read as an answer. "Base Layer 4471 Blue M" gets read as a warehouse label. Write the title the way a buyer describes the product out loud, following the structure in the Google Merchant product data specification. Type, key attribute, defining spec. In that order.
Description is where the recommendation gets made
The title gets you shortlisted. The description gets you picked. This is where the model finds the details it needs to say why your product fits.
Lead with what the thing is and who it's for. Then the material, the size range, the use case. Skip the brand poetry. An assistant quoting your product wants facts it can repeat, not adjectives it has to ignore.
Price and availability are the silent disqualifiers
Price and availability don't win the recommendation, but they lose it fast. If your price is missing or your availability field says out of stock, the model drops you before it weighs anything else.
Keep both accurate and current, and mark them up with a valid Offer object. An assistant will not recommend a product it can't confirm is buyable at a stated price.
Fix the feed in the order the model reads it
Don't start with GTINs and shipping weights. Start with title, then description, then price and availability. That's the order the machine reads, so that's the order you fix.
Run one product through the question you want to win. If the top four fields answer it clearly, you're in the game. Then move down the feed and clean the structured data, following Google's product structured data guidelines.
Further reading
- Google Merchant product data specification
- Schema.org Product reference
- Google product structured data guidelines
- Schema.org Offer reference
Frequently Asked Questions
Which product feed field matters most for AI search?
The title. It's the first field an AI assistant reads to decide if your product matches the shopper's question. A clear, descriptive title gets you shortlisted; a SKU-style title gets you skipped.
Do AI assistants read the whole product feed?
No. They pattern-match against the shopper's query and read the top fields first: title, description, price, and availability. Lower-priority fields like GTIN and shipping weight act as tiebreakers, not deciders.
Does structured data change which fields AI reads?
Structured data confirms and standardizes the fields, which makes them easier to trust. Mark up price and availability with a valid Offer object so an assistant can verify the product is buyable before recommending it.
Want to know if AI assistants can actually find and recommend your store? Get a free AI-visibility read on your Shopify store at WRKNG Digital. We show you exactly what ChatGPT, Perplexity, and Google AI see when they look at your products.
