Your Shopify Store Has More Product Data Than AI Can See

Good ideas travel faster when you share them; forward this to someone who’d find it useful.

We expected to find missing product data.

We'd spent about six weeks building and measuring AI visibility for Flaunt Your Ink (FYI), a D2C tattoo aftercare brand running on Shopify. The brand sells 33 active SKUs across three lines — tattoo aftercare, men's grooming, and merchandise. Standard Shopify setup. Google & YouTube channel connected. An agency managing paid Shopping campaigns. The founder handling theme edits via Shopify CLI. Everybody doing their job.

When we assessed how the brand showed up across AI shopping surfaces, we assumed the main problem would be gaps. Missing descriptions. Empty fields. Products that hadn't been listed.

What we found was different. The product catalog knew more about every SKU than any surface representing those products was actually exposing.

The problem wasn’t missing data. It was disconnected data.

A note on what “see” means throughout this piece: we mean available through the external representations we could inspect — storefront HTML, Product JSON-LD (the structured data embedded in product pages that search and AI systems read) and downstream commerce feeds. We cannot observe what any individual AI model does with every field it receives. What we can measure is whether the information reaches those surfaces at all.

Shopify gives merchants a stronger starting point than many commerce platforms.

Your product data lives in named, structured fields — title, description, product type, vendor, tags, variants, images, price, inventory, weight. The Google & YouTube channel integration handles a significant amount of plumbing to get your catalog into Google’s commerce ecosystem. Many themes emit Product structured data on product pages automatically, though the exact markup depends on the theme.

On FYI, the baseline was solid. The theme’s JSON-LD included product name, full description, brand, images, price, availability, review ratings, SKU, and GTIN for products that had barcodes. Google’s Rich Results Test confirmed a valid Product snippet with all the standard fields.

The plumbing works. The question is what’s happening beyond the plumbing.

Traditional search largely asks whether a page can be discovered, crawled, indexed, and ranked. Product discovery through AI systems introduces a different representation problem: does the system have enough structured, trustworthy product information to understand, compare, and represent the product inside an answer? In an AI shopping surface, the product itself may become the unit being represented — as a product card, recommendation, or comparison. A product can therefore be indexed without all of its available information being exposed through the representations downstream systems receive.

Shopify has a system called metafields — structured attributes attached to products beyond the default fields. Think of them as custom columns in a spreadsheet behind each product. Ingredients, materials, certifications, skin type compatibility — anything the standard product form doesn’t have a slot for.

On FYI, we opened the metafield layer on a single SKU — the Daily Tattoo Moisturiser — and found three distinct groups of structured data that most merchants don’t know are there.

The first group was Shopify’s own taxonomy. These fields are part of Shopify’s product taxonomy system and are populated based on the product’s category assignment. For the Daily Tattoo Moisturiser, classified as “Lotions & Moisturizers in Skin Care,” the system had populated fields like cosmetic function, product certifications, product form, skin care effect, suitable skin type, and age group. Shopify had even queued five additional suggestions the merchant hadn’t reviewed. These fields live in the backend. The merchant may not know they’re there.

The second group was merchant-created. The founder had built custom metafields for the theme: ingredient feature cards (Sodium PCA, Hyaluronic Acid, Betaine, Aloe Vera — each with a title, image, and description), benefit scroller text, and product bundle contents. This is where the richest product knowledge lived — real ingredient descriptions, real benefit claims, written by the person who formulated the product.

The third group was empty scaffolding. Four Google Shopping-related variant fields existed — Age Group, Condition, Gender, MPN — all blank. We couldn’t determine from the audit alone which integration had created the definitions.

The result was three layers of product knowledge that had accumulated independently. Shopify knew about some of the attributes. The founder had entered others. The Google-related definitions existed but were empty. Nothing in the Admin made those relationships obvious, and there was no single view showing what ultimately reached each downstream system.

Data in your Shopify Admin and data in your product’s structured markup are two different things.

On FYI, we compared what was in the Admin metafields against what the theme was actually emitting as Product JSON-LD.