By Steve Merrill, Founder of WRKNG Digital — October 20, 2026
What structured data do Shopify stores need to be recommended by AI?
Shopify stores need valid Product schema on every product page, with price, availability, brand, a product identifier (GTIN or MPN), and aggregated reviews nested inside it. That is the data an AI assistant reads to decide whether to recommend you. Without it, you are a guess.
Here's the thing. AI assistants don't shop the way people do. They can't squint at your hero image or feel out your brand. They parse. They pull the fields they can trust and skip the ones they can't read.
I've run this check on dozens of Shopify stores doing between $500K and $10M a year. Most of them think they're covered because the theme spit out a little schema. They're not.
Why does structured data decide whether AI recommends you?
An AI assistant recommends products it can describe accurately. Structured data hands it the facts in a format it trusts. Plain HTML forces the model to guess, and models hedge when they guess.
Think about what Perplexity or ChatGPT does when someone asks for "the best merino base layer under $120." It needs a name, a price, a stock status, and some proof the thing is good. If your page states all of that in machine-readable JSON-LD, you're a candidate. If it's buried in a carousel and a review widget that loads with JavaScript, you're invisible.
Schema.org's Product type is the shared vocabulary every major engine agreed to use. Google Search Central's product structured data docs spell out which fields are required versus recommended. These aren't WRKNG opinions. They're the published rules.
Which fields actually matter?
Six fields do most of the work: name, price, priceCurrency, availability, a product identifier, and aggregateRating. Get those right and an AI assistant can quote your product in an answer. Miss them and it moves on to a competitor who filled them in.
Here's what each one does for the machine reading your page.
- name and description. Tells the model what the product is in words it can repeat back.
- offers (price, priceCurrency, availability). The deciding facts for "under $120" or "in stock" queries.
- brand. Lets the model group you correctly and trust the listing.
- gtin or mpn. The barcode. Ties your product to the same item everywhere else on the web.
- aggregateRating and review. Social proof the model uses to rank you against similar products.
- image. A clean, direct image URL the assistant can surface.
Shopify's own theme SEO metadata docs show where this lives in your Liquid templates. The schema usually sits in a snippet your product template renders. That's the file you'll be editing.
One field people skip: aggregateRating. I audited a home goods store last quarter that had perfect price and availability schema and zero review data. AI tools kept recommending a weaker competitor with 40 reviews marked up. The better product lost because the model couldn't see the proof.
How do you add Product structured data to a Shopify store?
You add or fix the JSON-LD in your product template, make it match what shoppers see, then validate it. Four steps. None of them require a developer if your theme is reasonably modern.
- Audit what you have. Take three product URLs and run each through Google's Rich Results Test. It shows the exact Product fields your theme outputs and flags what's missing.
- Fill the gaps. Open your product template (often
product.liquidor a schema snippet) and add the missing fields. Pull live values from Shopify objects likeproduct.price,product.available, and your reviews app so the data stays accurate as inventory changes. - Match the page. The price, stock status, and rating in your JSON-LD have to match what's visible on the page. Google ignores schema that contradicts the content, and so do the assistants pulling from it.
- Validate and resubmit. Re-run the Rich Results Test until it's clean, then request indexing in Google Search Console so the recrawl happens sooner.
If you use a reviews app like Judge.me or Okendo, confirm it injects aggregateRating into your Product schema, not just into a visual widget. A lot of them only do the widget by default. That's a setting, not a rewrite.
What gets your structured data ignored?
Mismatched data is the biggest killer. If your schema says $89 and the page says $99, the engine throws the markup out. Same with fake or stale ratings. The model needs the JSON-LD and the visible page to tell one story.
Two more traps I see constantly. Review markup on category pages instead of product pages. And prices that only render in JavaScript after the schema has already been read. Both make you look empty to a crawler.
Blank fields count against you too. A Product object with name and nothing else is weaker than no schema at all, because it signals a half-built page.
One more thing on timing. People add the schema, see nothing in a week, and rip it back out. Don't. The crawl has to happen, the index has to update, and the assistants have to pull the fresh data. I've watched a store go from zero AI mentions to showing up in Perplexity answers three weeks after a clean fix. The work was done on day one. The payoff showed up on day twenty-one.
Frequently asked questions
Does Shopify add structured data automatically?
Most Shopify themes add some Product schema, but it's often incomplete. Many skip gtin, brand, and aggregateRating, which are the fields AI assistants rely on most. Check your own pages instead of assuming the theme handled it.
Which schema type matters most for AI product recommendations?
Product schema with Offer and AggregateRating nested inside it. That combination gives an assistant the price, availability, identifier, and social proof it needs to recommend your item with confidence.
Do I need a GTIN for every product?
If your products carry manufacturer barcodes, yes. GTINs let AI tools match your listing to the same product across the web. For handmade or private-label goods with no GTIN, use brand plus mpn instead.
How long until AI assistants pick up my structured data?
Google usually recrawls within days to a few weeks. Assistants that pull from live search results reflect changes faster. Give it two to four weeks before you judge the impact.
Where to go from here
Structured data is the floor, not the ceiling. It gets you readable. Getting recommended means your feed, your reviews, and your page content all line up with what shoppers ask AI every day.
That's the work we do. If you want your Shopify store built to be recommended by AI, not just indexed by it, see how WRKNG Digital approaches agentic commerce.
