Open a live product page on your store, view its page source and search for application/ld+json before you add any code. Many Shopify themes already output structured data, so the first job is to see what is there and whether it is accurate.

From there the path is short. Duplicate your theme, decide between a small Liquid edit and a tested app, keep one accurate Product graph per product page, and validate it with Google's Rich Results Test and the Schema Markup Validator. Bing and Yandex offer their own validators when those engines matter to your store.

Key takeaways

  1. Inspect before addingView a live product page's source and search for application/ld+json. Your theme or an app may already output a Product graph.
  2. Duplicate the themeShopify advises duplicating a theme before code changes, because theme updates can remove customizations that do not fit the new version.
  3. One graph per productKeep one accurate Product graph per product page, using the price, currency, availability, image, URL and SKU that visitors can see.
  4. Variants and reviewsUse ProductGroup when variants have their own URLs, and add reviews or ratings only when they are genuine, visible and about that product.
  5. Validate, then monitorTest with the Rich Results Test and Schema Markup Validator, then watch Search Console. Valid markup does not guarantee a rich result, ranking or AI citation.

Check what your store already outputs

Your store may already describe its products to search engines, so look before you write anything. Shopify themes are built from templates and sections, and a section can print a JSON-LD script into the page. Whether yours does depends on the theme, its version, your product data and any apps you have installed.

A man in a checked shirt reads a laptop at a packing table in a small soap workshop, with wrapped soap bars beside him
Figure 1. Viewing a live product page's source is the first step, before any code is added.

Do this on the live store rather than in the theme editor preview, because the preview can hide what visitors and crawlers receive:

  1. Open a product page in your browser.
  2. View the page source and search for application/ld+json.
  3. Count the blocks and read the @type of each one. Look for Product, ProductGroup, Organization and WebSite.
  4. Compare the price, currency, availability, image, URL and SKU in the markup with what the page shows.

Apps can inject markup too, so inspect the rendered page in your browser's developer tools as well as the raw source. If you find two Product blocks, or one block whose price differs from the visible price, note it now. The duplicate section below explains how to resolve it.

What Dawn and Online Store 2.0 output

JSON-LD is a machine-readable format for describing information that is visible on a page. Schema.org supplies the vocabulary, such as Product and Offer, while Google, Bing, Yandex and other systems decide which properties they use (JSON-LD 1.1, W3C).

Online Store 2.0 themes use JSON templates as wrappers around Liquid sections, and Liquid files generate the HTML and can contain JSON-LD scripts, according to Shopify's theme architecture documentation. Shopify's Dawn reference theme shows what that looks like. Its header section outputs an Organization object with the shop name, logo, social profiles and URL, and its homepage output includes a WebSite object with a SearchAction. Its main product section contains the product JSON-LD.

So, does Shopify add schema markup automatically? Dawn does, but the Dawn repository changes over time, and customized or older themes can differ. Your installed theme is the reference for your store, which is why the inspection comes first.

The SearchAction in that WebSite object deserves one note. Google retired the sitelinks search box (Google Search Central blog, June 2025), so adding SearchAction only to chase that display is no longer a reason. The vocabulary itself remains valid Schema.org, and other consumers can still read it, so there is no need to hurry to remove what your theme already outputs.

Choose theme Liquid or an app

Editing theme Liquid suits a store with a developer or a technically confident owner. It gives you maximum control, works for a precise correction to existing markup, and handles custom variant, brand or offer logic without another app dependency. The cost is upkeep: Shopify warns that theme updates can remove customizations that are incompatible with the new version, so keep a note of every edit you make.

An app suits a store that does not want to maintain Liquid. Look for an app that documents exactly what markup it adds, updates its data from your Shopify products and variants, and can disable overlapping markup or recognize what the theme already outputs. Apps can also duplicate markup, conflict with each other, depend on a subscription or leave stale data behind, so test any app on a duplicate theme first.

A page card splitting into a hand-edit path and a plug-in path that rejoin at one glowing product card with a check mark
Figure 2. Liquid edits and apps are two routes to the same goal, one accurate product graph.

A Custom Liquid section sits between the two. It is useful for a controlled addition, but it is easy to place markup in the wrong template or to repeat a graph that already exists. Whichever route you pick, Shopify advises duplicating the theme before code changes, and the duplicate is where you test.

If you also run a WordPress site, the WordPress schema markup guide follows the same check-first routine for plugins and themes.

Add or fix one Product graph in Liquid

Edit only when the current output is missing or inaccurate. In Dawn the product JSON-LD lives in sections/main-product.liquid, though other themes use a snippet or a section rendered on product templates. The exact file varies by theme. If your theme's block is accurate, leave it alone; if it is missing something, merge the missing property into that block rather than adding a second one.

Google says Product markup can make a page eligible for product snippets or merchant listings, and that eligibility does not guarantee a display (Google documentation updated September 2026). Merchant listings need a name, an image and a purchasable Offer with a price and currency. Product snippets need a name and at least one of offers, review or aggregateRating.

This fictional single product shows a complete graph for one product page:

json
{
  "@context": "https://schema.org",
  "@type": "Product",
  "@id": "https://example.com/products/blue-harbor-lantern#product",
  "url": "https://example.com/products/blue-harbor-lantern",
  "name": "Blue Harbor Lantern",
  "image": [
    "https://example.com/images/blue-harbor-lantern.jpg"
  ],
  "description": "A fictional rechargeable lantern for indoor and outdoor use.",
  "sku": "EX-LANTERN-001",
  "brand": {
    "@type": "Brand",
    "name": "Example Workshop"
  },
  "offers": {
    "@type": "Offer",
    "url": "https://example.com/products/blue-harbor-lantern",
    "priceCurrency": "USD",
    "price": 39.95,
    "availability": "https://schema.org/InStock",
    "itemCondition": "https://schema.org/NewCondition"
  }
}

Replace every value with data a visitor can see on that product page, and fill each one from the matching value in your product data the way your theme's existing block does. In a theme file this JSON sits inside a script element with type="application/ld+json". Keep the JSON strict, with no comments and no trailing commas.

To build the same graph without hand-writing it, use the Product schema generator and compare its output with your theme's block. The Product schema markup guide explains each property in more depth.

An Organization object for your store is a separate, optional decision. Google has no product rich-result requirement for it. Add one only when it accurately describes your visible business identity, and list only official profiles in sameAs.

Google requires some of these properties for merchant listings and only recommends others, and a few are optional. Schema.org permits a broader vocabulary than Google currently uses, so a property can be valid without being required for this Google feature.

Product properties by status for Google merchant listings

Hover over or tap a row to highlight it. On small screens the table scrolls sideways.

PropertyStatus for this Google featureWhat to enter
name, image, offersRequiredA visible product name, an image URL and a purchasable Offer
price, priceCurrencyRequiredThe price shown on the page and its currency code
Offer url, availability, itemConditionRecommendedThe product URL, a Schema.org availability value and the item condition
brandRecommended when knownThe brand name, as a Brand object
description, skuOptionalA short visible description and the SKU shown for the product
aggregateRating, reviewOnly when genuineRatings or reviews that are genuine, visible and about this product

Mark up variants with ProductGroup

A product page can represent one product with several variants, such as colors or sizes. Google supports ProductGroup with hasVariant, variesBy and productGroupID to describe how those variants relate (Google's product variants announcement, February 2024).

This is the pattern when each variant has its own URL. Each variant carries its own name, SKU, URL, image, price, currency and availability wherever those values differ:

json
{
  "@context": "https://schema.org",
  "@type": "ProductGroup",
  "@id": "https://example.com/products/blue-harbor-lantern#group",
  "name": "Blue Harbor Lantern",
  "description": "A fictional rechargeable lantern sold in two colors.",
  "productGroupID": "EX-LANTERN",
  "variesBy": [
    "https://schema.org/color"
  ],
  "hasVariant": [
    {
      "@type": "Product",
      "@id": "https://example.com/products/blue-harbor-lantern?variant=blue#product",
      "isVariantOf": {
        "@id": "https://example.com/products/blue-harbor-lantern#group"
      },
      "url": "https://example.com/products/blue-harbor-lantern?variant=blue",
      "name": "Blue Harbor Lantern, Blue",
      "color": "Blue",
      "image": [
        "https://example.com/images/blue-harbor-lantern-blue.jpg"
      ],
      "sku": "EX-LANTERN-BLUE",
      "offers": {
        "@type": "Offer",
        "url": "https://example.com/products/blue-harbor-lantern?variant=blue",
        "priceCurrency": "USD",
        "price": 39.95,
        "availability": "https://schema.org/InStock",
        "itemCondition": "https://schema.org/NewCondition"
      }
    },
    {
      "@type": "Product",
      "@id": "https://example.com/products/blue-harbor-lantern?variant=green#product",
      "isVariantOf": {
        "@id": "https://example.com/products/blue-harbor-lantern#group"
      },
      "url": "https://example.com/products/blue-harbor-lantern?variant=green",
      "name": "Blue Harbor Lantern, Green",
      "color": "Green",
      "image": [
        "https://example.com/images/blue-harbor-lantern-green.jpg"
      ],
      "sku": "EX-LANTERN-GREEN",
      "offers": {
        "@type": "Offer",
        "url": "https://example.com/products/blue-harbor-lantern?variant=green",
        "priceCurrency": "USD",
        "price": 41.95,
        "availability": "https://schema.org/OutOfStock",
        "itemCondition": "https://schema.org/NewCondition"
      }
    }
  ]
}

Change the group name and productGroupID to match your product, then give each variant the URL, SKU, image and offer it really has. The green variant here has a different price and is out of stock, which is the kind of difference that belongs on the variant and not on the group. For Google's variant feature, ProductGroup needs a name, productGroupID, variesBy and hasVariant, and each variant's color or other varying property needs a value.

If variants have distinct URLs, each URL must identify its own variant accurately. Do not describe every variant as an unrelated product unless the page and URL support that reading. The product variant schema guide covers the one-page and many-pages cases in more detail.

Show reviews and ratings only when they are genuine

Add Review or AggregateRating only when the review or rating is genuine, visible on the page and about the product you marked up. Never invent ratings, review counts or reviewer names. A customer review on your own product page is also a different thing from an editorial product review, and Google says its pros-and-cons display applies to editorial review pages, not to ordinary merchant product pages or customer reviews.

Review apps often add both the visible review widget and its JSON-LD. Before you write a review graph by hand, inspect the rendered page to see whether the app already outputs one. If it does, check that it describes this product and that its numbers match what visitors see. The review schema markup guide explains the permitted contexts.

Remove duplicate markup

Two or more graphs describing the same product can leave a page with conflicting data, for example two prices or two availability values. Duplicates usually come from one of these situations:

  • Dawn already outputs a product graph and an app adds another.
  • A theme snippet is included both globally and on the product template.
  • A review app adds a second Product object with a different price or availability.
  • A product graph is added to collection cards instead of product pages.
  • Two theme versions stay active in preview and get confused during testing.

The preferred fix is to keep one authoritative product graph, merge any missing supported properties into it and remove or disable the overlapping output. Decide which graph is authoritative by asking which one you can keep accurate over time. Do not delete existing markup until you have validated its replacement, so a product page is never left without any graph.

When Search Console or a validator reports missing fields, the guide to fixing missing field schema warnings walks through the common causes.

Validate and monitor

Use several layers, because each answers a different question. A parser tells you the JSON is well formed, Schema.org vocabulary checks tell you the types and properties are valid, and a Google feature test tells you whether the page can be eligible for that feature. None of them tells you whether the result is indexed or shown.

Start with your live page and work through these checks:

  1. Inspect the page source and rendered HTML for application/ld+json.
  2. Confirm the JSON parses, with no comments or trailing commas.
  3. Run the URL through Google's Rich Results Test for Google feature eligibility.
  4. Run it through the Schema Markup Validator for general Schema.org vocabulary checks, or paste code into the Schema Nova validator.
  5. After the page is indexed, read the Merchant listings and Product snippets reports in Google Search Console, where Google says errors and warnings can appear.
  6. Recheck after theme updates, review-app changes, product-feed changes or price changes.

Bing offers a Markup Validator in Bing Webmaster Tools and says it supports JSON-LD and Schema.org; see also the Bing Webmaster Blog's JSON-LD announcement. The Yandex Structured Data Validator recognizes Schema.org, JSON-LD, Microdata, RDFa, Open Graph and microformats. The structured data testing tool replacement guide explains how these tools divide the work.

Setup steps (file location varies by theme)

  1. Inspect existing markupView a live product page's source and search for application/ld+json.
  2. Duplicate the themeMake the changes on the copy, not the live theme.
  3. Choose Liquid or an appPick the route you can maintain, and confirm what an app adds.
  4. Add one authoritative graphFix or add one Product graph using visible data, and remove overlaps.
  5. Test syntax and eligibilityRun the Schema Markup Validator and Google's Rich Results Test.
  6. Deploy and monitorPublish, then watch Search Console reports after indexing.

Bing says annotations may supplement or validate its data and may improve how results look, and DuckDuckGo results may include data from Bing, so the same accurate Product JSON-LD serves them. Yandex says Schema.org markup can support product and price snippets (Yandex Schema.org guidance). Google says AI Overviews and AI Mode need no special AI markup, and that structured data should match the visible content (Google AI features documentation, updated December 2025).

OpenAI recommends allowing OAI-SearchBot in robots.txt if you want inclusion in ChatGPT search (OpenAI crawler documentation), and Perplexity recommends reviewing robots.txt and firewall rules for PerplexityBot (Perplexity crawler guidance). Keep product facts visible and canonical URLs consistent. See whether AI systems use schema markup.

Questions people ask

Does Shopify add schema markup automatically?

Many themes do. Dawn outputs Organization and WebSite data in its header and product JSON-LD in its product section, but the output depends on your theme version, product data and apps. Inspect a live product page to see your own store's result.

Should I add a schema app if my theme already has JSON-LD?

Only if the app covers something your theme's output lacks and can avoid a second Product graph. Ask the app to identify or disable overlapping output, and test it on a duplicate theme before it goes live.

Should I use Product or ProductGroup for variants?

Use a plain Product for a single item. Use ProductGroup with hasVariant, variesBy and productGroupID when a product has variants such as color or size, and give each variant that has its own URL accurate details for that URL.

Can a review app add review schema?

Yes, and many apps output both the visible reviews and the markup. Inspect the rendered page first, and keep the markup only if the reviews are genuine, visible and about that product.

Does valid schema guarantee rich results?

No. Google says eligibility is not a guarantee of display, and Bing says the same about rich snippets. Valid markup also does not guarantee rankings, traffic or citations in AI answers.

Quick check

Choose an answer to see feedback. Nothing is scored or saved.

Question 1 of 3

A JSON-LD block parses correctly but describes a product that is not visible on the page. Is it safe to publish?

Question 2 of 3

A review app and your theme each output a Product graph. What should you do?

Question 3 of 3

The Rich Results Test shows a product as eligible. What does that mean?

Glossary

Select a term to see its definition.

JSON-LD. A JSON-based format for linked, machine-readable data.