By Steve Merrill, Founder of WRKNG Digital | June 5, 2026
Before ChatGPT Shopping, Perplexity, or Google AI Mode recommends your product, it runs a quiet checklist. Most Shopify stores fail at least half of it. Here are the ten things AI shopping agents actually evaluate, and what passing each one looks like.
1. Product Title Structure
Your title is the first signal an AI agent parses, and a vague title ends the evaluation before it begins. According to Google's Merchant Center product data specifications, titles should follow a structured order: brand, product type, then attributes like size, color, or material. "Blue Sneakers" fails. "Nike Men's Air Zoom Pegasus 41, Blue, Size 10" matches buyer intent and gets recommended.
2. GTIN or Unique Product Identifier
The Global Trade Item Number is how AI agents verify they have found the right product across multiple data sources. Google requires a GTIN for any product a manufacturer has assigned one to, and ChatGPT Shopping uses the same identifiers to cross-reference listings before surfacing a recommendation. No GTIN means no cross-referencing. No cross-referencing means no recommendation. Most Shopify stores leave this field completely blank.
3. Natural Language Product Description
AI shopping agents use your description to answer buyer questions, not to display it verbatim. Write the way a knowledgeable salesperson talks: materials, fit, use case, who the product is for. Marketing copy built for humans rarely passes AI readiness criteria. Google recommends 500 to 1,000 characters for most product types, with up to 5,000 available for complex items.
4. Live Price and Availability
AI shopping agents verify price and stock before recommending anything. Perplexity's shopping features and ChatGPT Shopping both pull live pricing data before surfacing a result. Stale data does not get displayed, it gets filtered out entirely. Stores with active inventory need feed updates every 30 to 60 minutes, or a sync trigger whenever inventory changes.
5. On-Page Product Schema Markup
Your product feed covers what lives in your data pipeline. Product schema covers what AI crawlers read directly off your page. Schema.org's Product markup includes fields for name, brand, SKU, price, currency, availability, and aggregate ratings. Google AI Mode reads schema before it reads your page copy. If schema says "out of stock" and the page says "buy now," AI defers to the schema every time.
6. Aggregate Ratings and Review Count
AI shopping agents treat review data as a trust signal, not a vanity metric. A product with 4.6 stars from 280 reviews gets recommended ahead of a product with 5 stars from 3. The rating needs to be structured in your Product schema using the aggregateRating property, not just displayed visually on the page. If your reviews live in Shopify but are not in your schema, the AI cannot read them.
7. Brand Field Consistency
AI agents cross-reference your brand field against known brand registries and structured data from your website. A brand listed as "Nike" in your feed but "nike_us" on your product schema creates a mismatch that lowers recommendation confidence. Your brand name needs to be identical, correctly capitalized, and consistent everywhere it appears, feed, schema, title, and page content. This is an easy fix most stores never make.
8. Google Product Category Specificity
Google's Product Category taxonomy has over 6,000 categories. Picking a precise one tells AI agents exactly where your product belongs in a recommendation set. "Apparel" is not a category. "Apparel & Accessories > Clothing > Activewear > Running Pants" is. The more specific the category, the better the product performs across AI-powered shopping surfaces.
9. Shipping and Return Policy Signals
Buyer intent questions often include shipping or return conditions. "Free shipping running shoes" and "easy return policy boots" are real queries AI shopping agents handle. If your shipping and return data is not structured in your feed or schema, AI cannot match your product to those queries. Shopify stores can expose this data through the shippingDetails and hasMerchantReturnPolicy schema properties, most do not.
10. Image Count and Quality
Multimodal AI shopping agents now process product images directly, not just the text around them. Your primary image needs a clean background and a descriptive alt text attribute that includes the product name and key attributes. Most Shopify stores submit one image. The stores getting recommended submit four to six, lifestyle shots, detail views, and scale references that give AI agents visual context to match a buyer's search intent.
How We Built This List
This list comes from AI readiness audits of Shopify product catalogs, matched against published data requirements for ChatGPT Shopping, Perplexity, and Google AI Mode. Each factor was weighted by how often its absence caused a product to be excluded from a recommendation set entirely, before any other signal was even evaluated.
FAQ
Which of these checks do most Shopify stores fail?
GTIN, on-page Product schema, and image count. These three gaps alone exclude most Shopify products from AI recommendation sets before any other signal is evaluated. They are also among the easiest to fix once you know they are missing.
Do I need to submit a product feed to appear in AI shopping recommendations?
For ChatGPT Shopping and Google AI Mode, yes, a merchant feed submission is required for commercial product recommendations. On-page schema helps AI crawlers find and understand your products, but it does not replace feed submission for commercial placements.
How do AI shopping agents handle out-of-stock products?
They filter them out. An AI agent will not recommend a product it cannot confirm is available at the listed price. Stale availability data is one of the most common reasons products disappear from AI recommendation sets without any obvious explanation.
Does having more reviews automatically improve AI shopping visibility?
Volume helps, but structure matters more. Your review data must be in Product schema using the aggregateRating property. Reviews visible on your page but absent from your schema are invisible to AI agents that parse structured data before reading page content.
How long does it take to fix these issues on a Shopify store?
Simple fixes like brand field consistency and image alt text can be done in a day. Feed submission, GTIN enrichment, and full Product schema implementation typically take one to two weeks depending on catalog size. The stores that act now are the ones that show up when AI shopping traffic accelerates.
Want to see exactly which of these checks your Shopify store is failing right now? Get a free AI commerce readiness audit at wrkngdigital.com/agentic-commerce-landing-page. We run every product through the same criteria AI shopping agents use and show you exactly what is costing you recommendations.
