AI engines do not read your marketing copy. They read your structured data. Here is exactly which schema types a Shopify store needs, how to check what your theme actually outputs, and the two traps that quietly break it, from the stores we run ourselves.
A Shopify store needs, at minimum, Product and Offer schema on every product page, Review and AggregateRating where the ratings are genuine, Organization schema for the brand, and FAQPage where you answer buyer questions. That is the machine-readable layer AI engines actually read when they decide whether to name your store in an answer. The catch is that your theme probably ships some of it already, incomplete or slightly wrong, and you cannot see the gaps without testing. We run two Shopify stores of our own, Pet Buckets and EMPRICS, so what follows is the exact schema pass we run on our own storefronts before a client ever sees it.
Product schema is structured data: a block of machine-readable facts about a product, written in the shared vocabulary at schema.org and embedded in the page, usually as JSON-LD. It states the name, the price, the currency, whether the item is in stock, and the rating, in a format a machine can parse without guessing from your layout. A human reads your product page. An engine reads the schema behind it.
This matters more in the AI era than it did for classic search, because a generative engine is not just ranking your link, it is describing your product in a sentence. To do that with confidence it wants facts it can lift cleanly and cross-check, not marketing prose it has to interpret. Structured data is the cleanest way to hand it those facts. It is not a ranking trick, it is you removing the ambiguity that would otherwise make an engine reach for a competitor it understands better.
The same markup does double duty. It is what powers rich results in Google search, and it is part of what merchant and AI shopping surfaces read to understand your catalogue. One correct block of JSON-LD, many readers.
Fewer than the internet suggests, and each with a clear job. Here is the working set, what each one tells an engine, and where it belongs.
| Schema type | What it tells the engine | Where it lives |
|---|---|---|
| Product + Offer | Name, price, currency, availability, condition | Every product page |
| Review + AggregateRating | Genuine ratings it can corroborate | Product pages with real reviews |
| Organization | Who the brand is, as one consistent entity | Site-wide, usually the homepage |
| FAQPage | Question-and-answer pairs it can quote | Product, collection and guide pages |
| BreadcrumbList | Where a page sits in the site structure | Collections and articles |
Product and Offer are the non-negotiable pair, and Offer is where stores most often fall short. An Offer with a name but no price, currency or availability is a half-answer, and a half-answer is exactly what makes an engine hesitate. Organization schema is the piece most stores skip entirely, yet it is what lets an engine treat your brand as a real, single entity rather than an unknown domain, which is the foundation the whole recommendation sits on.
One honest note on FAQPage. In 2023 Google narrowed FAQ rich results to authoritative government and health sites, so most stores will not see the old dropdown in search. The markup is still worth keeping, because it structures your answers cleanly for AI engines and other parsers, but do not add it expecting the search feature back. Mark up what genuinely helps a machine read the page, not what chases a result that no longer shows.
Receipt · Client EMPRICS · 900+ SKUs on ShopifyA specification-grade B2B catalogue where product data structure was the whole job. This is Product and Offer schema at scale. Open →You test a live URL, because guessing is where stores go wrong. Paste a real product page into Google's Rich Results Test, or the Schema Markup Validator, and read back exactly which types are detected and which fields it flags as missing or invalid. That report is the truth. Your theme's marketing page is not.
Do it per template, not once for the whole store. Shopify renders schema at the template level, so a product page, a collection page and the homepage can each output something different. Test one of each. On our own stores we check a product URL for Product and Offer, the homepage for Organization, and a collection for BreadcrumbList, then fix whatever the validator flags before moving on.
Where does the markup live in Shopify? Usually inside the theme's JSON-LD, often in a snippet the theme injects, sometimes added or overridden by an app or a metafield-driven block. You do not need to hand-code it to audit it. You do need to read the validator honestly and resist the urge to assume the theme has it covered, because in our experience it usually does not, fully.
The first is partial or wrong product schema shipped by the theme. Many Shopify themes output a Product block that is missing availability, missing currency, or carries a stale price, and it validates as present while being incomplete. Present is not the same as correct. The only way to know is the test in the section above, and the fix is to complete the fields the validator flags rather than trusting the default.
The second is the canonical trap. Shopify auto-applies canonical tags so that a collection-scoped product URL, the /collections/x/products/y form, points back to the base /products/y version. That is correct behaviour and you should leave it alone. What it means for you is that your schema, your internal links and any feed should reference the canonical base product URL, so engines consolidate every signal on one page instead of splitting it across duplicates. Fight the canonical and you scatter your own authority. Work with it and it does you a favour.
Never add Review or AggregateRating schema for ratings you do not genuinely have and show on the page. It is tempting, because a star rating in an answer is persuasive, and it is exactly the shortcut that gets a store a Google manual action. Google's structured data guidelines are explicit that markup must reflect content actually visible to the user, and fabricated review markup is treated as spam.
The deeper reason is the one this whole site is built on. Schema exists to help an engine trust you by giving it facts it can verify. Marking up a rating you invented is the precise opposite of that, and if it is caught it burns the trust you were trying to build. Mark up the reviews you really have. When you have none yet, ship Product and Offer cleanly and earn the reviews first. Receipts, not props.
The order we work in on Pet Buckets and EMPRICS, top to bottom, because sequence keeps you from polishing the wrong thing first.
We will run your product pages through the validator, show you exactly what your theme is missing, and fix the Product, Offer and Organization markup so an AI engine can actually trust it. On our own stores, then on yours.
Get a schema audit→no retainer trap. just truth.