guide · Product data

Product schema markup: match every marked-up value to the visible page

Product markup is a promise that a machine-readable copy says what the page says. Here is an annotated example, the checks that keep it honest, and what the evidence says about AI answers.

The short answer

Mark up each product page with Product and a single Offer whose name, price, currency, availability and identifiers match what a shopper can see on that page, then validate twice: once for syntax and Google's required properties, once by comparing every marked-up value with the visible text. Google says there is no special schema.org markup for its AI features, so treat markup as an exact copy of your page's facts.

Our take

Markup is worth doing for one reason: it hands machines an exact copy of your facts. If the markup and the page disagree, you've told every reader two different things.

Mark up each product page with Product and a single Offer whose name, price, currency, availability and identifiers match what a shopper can see on that page. Then validate twice. First check syntax and Google’s required properties. Then compare every marked-up value with the visible text, one by one.

Google says there’s “no special schema.org structured data” needed for its AI features, and it doesn’t guarantee rich results either. Rich results and AI choices are separate outcomes, and markup can’t promise either one.

What markup can do is put the same facts in a second, machine-readable place. Its job is to agree with the page.

Three outcomes people mix up

Most arguments about product schema blur three different questions.

Question	What the evidence says	Source
Is the markup valid and true?	Within your control. Google publishes required properties and says markup must describe content visible on the page	Google’s merchant listing and general guidelines pages
Will Google show a rich result?	Not guaranteed. Google says it “does not guarantee” structured data features will show, and may try to verify merchant data first	Google’s general guidelines and merchant listing pages
Will an AI assistant pick or cite the product?	Not documented as a markup effect. Google says no special markup is needed for AI features. One vendor study found adding JSON-LD barely moved AI citations on pages already cited	Google’s AI features page; the Ahrefs study

Keep that third row honest in both directions. Ahrefs tracked 1,885 pages that added JSON-LD and compared them with matched control pages. Adding JSON-LD produced “no major uplift” in citations in AI Overviews, AI Mode or ChatGPT. But every page studied was already heavily cited, all schema types were pooled together, and the window was 30 days.

Ahrefs itself says there are still good reasons to use schema, rich results among them. A near-null citation result for pages that are already visible doesn’t show that Product markup is useless on a product page. It shows the markup isn’t a shortcut to AI citations.

On the shopping side, OpenAI says ChatGPT’s shopping results draw on “structured metadata from first-party and third-party providers (e.g., price, product description)”. It doesn’t say whether that includes JSON-LD on your page. Google’s Shopping help says its “Top recommendations” are selected based on relevance to the search, ratings, price and product features. Markup isn’t named as an input. Google says it combines structured data with Merchant Center feeds where both exist. The common thread is consistency. Several systems may hold copies of your facts, and the copies should agree. Our price and stock guide covers keeping page, feed and markup in step.

The annotated example

Hypothetical store and product, not a real listing. The page shows:

the title “Top-fill humidifier, 4 L, white”
“$69.99 was $89.99”
“In stock”
a one-line description
a link to the store’s returns policy

Here’s the Product markup for that page, with notes:

{
  "@context": "https://schema.org/",
  "@type": "Product",
  "name": "Top-fill humidifier, 4 L, white",
  "description": "Top-fill cool-mist humidifier with a 4 L tank.",
  "image": [
    "https://example.com/images/hum4-white-1x1.jpg",
    "https://example.com/images/hum4-white-4x3.jpg",
    "https://example.com/images/hum4-white-16x9.jpg"
  ],
  "sku": "HUM4-WHT",
  "brand": { "@type": "Brand", "name": "Example Home" },
  "offers": {
    "@type": "Offer",
    "url": "https://example.com/products/top-fill-humidifier?variant=white",
    "price": 69.99,
    "priceCurrency": "USD",
    "priceSpecification": {
      "@type": "UnitPriceSpecification",
      "priceType": "https://schema.org/StrikethroughPrice",
      "price": 89.99,
      "priceCurrency": "USD"
    },
    "availability": "https://schema.org/InStock",
    "itemCondition": "https://schema.org/NewCondition"
  }
}

Property	Note	Must match on the page	Rule source
`name`	Required. Same words as the visible title, including the variant	The H1 or product title	Google merchant listing: required properties
`image`	Required. Crawlable, and it must show the marked-up product. Google recommends several high-resolution images, in 16x9, 4x3 and 1x1	The product photos shown	Google merchant listing: image guidelines
`sku`	Your own ID. Fine to include	Not usually visible. It’s an internal code	Schema.org: a merchant-specific identifier
`gtin`	Left out on purpose. This product has no assigned GTIN	Nothing to match	Our identifier guide
`offers` `@type: Offer`	Required as a single `Offer`. An `AggregateOffer` doesn’t qualify for merchant listings	The purchase box	Google merchant listing: “the merchant has to be the seller”
`price`	The active price, above zero, as a number	“$69.99”	Google merchant listing: price
`priceSpecification` with `StrikethroughPrice`	The original price during a sale. Never mark the active price with a `priceType`	“was $89.99”	Google merchant listing: sale pricing
`priceCurrency`	A three-letter ISO 4217 code, never “$”	The currency shown	Google merchant listing: priceCurrency
`availability`	An `ItemAvailability` value	“In stock”	Google merchant listing: availability
`url`	The page for this variant	The URL the shopper lands on	Google merchant listing: Offer

Two things are missing on purpose. First, there’s no aggregateRating, because the page shows no reviews, and Google’s rules say not to mark up content readers can’t see. Second, there’s no return policy in this block. Google recommends a store-wide return policy nested under Organization markup, and product-level policies only to override it. Our shipping and returns markup guide covers that block.

For a product with several variants, schema.org’s ProductGroup represents “a group of Products that vary only in certain well-described ways, such as by size, color, material”. The group itself isn’t offered for sale. Its variants are. Our variant guide covers that structure.

The first draft, and what was wrong with it

A first draft for the same hypothetical page looked like this, shortened:

{ "@type": "Product", "name": "Top-Fill Humidifier 4L", "gtin": "HUM4-WHT",
  "offers": { "@type": "AggregateOffer", "price": "89.99",
              "priceCurrency": "$", "availability": "https://schema.org/InStock" } }

We ran both blocks through a short script that encodes a subset of Google’s stated rules and compares values with the page text. The corrected block passed. The draft failed in five places, plus one problem the script couldn’t see:

Problem	Why it matters	Fix
`AggregateOffer`	Merchant listings require a single `Offer`	`"@type": "Offer"`
`price` “89.99”	That’s the compare-at price. The active price the page shows is 69.99. Google lists price as a number	`69.99`, with 89.99 as `StrikethroughPrice`
`priceCurrency` “$”	Must be ISO 4217	`"USD"`
`gtin` “HUM4-WHT”	A SKU in the GTIN field fails the GTIN length and check-digit rules	Remove it and keep the value in `sku`
`name` doesn’t match the title	Markup should describe what readers see	Copy the visible title
Stock state	“In stock” is right today. If the template hard-codes it, the markup keeps saying it after the item sells out	Generate it from live inventory

The price row is the costly one. The draft would have told any system reading the markup that the product costs $89.99, while the page charges $69.99. That’s exactly the kind of mismatch Google’s guidelines warn about. It’s also the kind a shopper would notice.

The validation checklist

Syntax and required properties

The JSON-LD parses, with no trailing commas or smart quotes.
Product has name, image and offers.
offers is one Offer with a price above zero and an ISO 4217 currency.
A sale uses StrikethroughPrice on the original price only.
Run the page through Google’s Rich Results Test and fix any critical errors. Then use URL Inspection on a few deployed pages. Google’s page describes both steps.

Visible parity. Do this by hand or by script, on the rendered page:

name matches the visible title.
price matches the price the shopper pays. The strikethrough price matches the “was” price shown.
availability matches the stock message for the selected variant.
Every identifier is real. No SKU sits in gtin or mpn.
No rating, review or policy is marked up unless the page shows it.

Delivery

The markup is in the initial HTML. Google recommends this, and warns that JavaScript-generated markup can make Shopping crawls “less frequent and less reliable” for fast-changing price and availability.
Price and stock come from the same source as the visible page, not a cached template.
Markup, Merchant Center feed and page agree. Google says it may combine them, and may try to verify merchant listing data before showing it.

What markup can’t carry for you

Markup repeats facts. It doesn’t create them. If the page never says the humidifier’s tank size, runtime or noise level, there’s nothing true to mark up. And Google’s rule against marking up invisible content means you shouldn’t put those facts in JSON-LD alone.

Put the facts a shopper asks about on the page first, in words and with units. Then let markup copy them. Our requirement-by-requirement check shows how to find which requests a page leaves unanswered.

How we checked

On October 27, 2026 we read Google’s Introduction to Product structured data (last updated 2025-12-10), Merchant listing structured data (last updated 2026-09-08), General structured data guidelines (last updated 2026-07-10) and AI features pages. We also read schema.org’s ProductGroup and MerchantReturnPolicy definitions, OpenAI’s shopping help article, and the Ahrefs study (published 2026-05-11, modified 2026-06-09). Every rule in the annotation and checklist is quoted or closely paraphrased from those pages.

The draft and corrected blocks were checked with a short script. It parses the JSON and tests required Product properties, Offer versus AggregateOffer, a price above zero, the ISO currency, priceType usage, GTIN length and check digit, and whether the name, prices and stock text appear in the page text we wrote for the example. It isn’t the Rich Results Test, and it doesn’t replace Google’s own tools.

Your next step

Open one product page, view the source, and find the application/ld+json block. Put it side by side with the rendered page and fill in the “Must match on the page” column of the annotation table for every property it contains. Anything that doesn’t match, or that the page doesn’t show at all, is the first fix. Then test the page in the Rich Results Test.

Evidence behind each claim

Show all 9 claims with their sources, dates and limits

Google says you don't need new machine-readable files, AI text files or markup to appear in AI Overviews or AI Mode, that there is no special schema.org structured data to add, and lists making sure structured data matches the visible text on the page among SEO fundamentals that remain worthwhile.
Google Search Central: AI features and your website · 2026-10-27 · documented
Claim limitation: Covers Google Search AI features only. Page last updated 2025-12-10.

Google's general structured data guidelines say not to mark up content that is not visible to readers of the page, that a rich result may not show if the content the markup refers to is hidden from the user, and that Google does not guarantee structured data will show up in search results.
Google Search Central: General structured data guidelines · 2026-10-27 · documented
Claim limitation: Google's rules for its own rich results. Page last updated 2026-07-10.

Google's merchant listing page requires name, image and offers on Product; requires a nested Offer, not an AggregateOffer, because the merchant must be the seller; requires an active price greater than zero and an ISO 4217 priceCurrency; and marks a sale's original price with a priceType of StrikethroughPrice, never on the active price.
Google Search Central: Merchant listing (Product, Offer) structured data · 2026-10-27 · documented
Claim limitation: Requirements for Google merchant listing experiences only. Page last updated 2026-09-08.

The same page says only pages where a shopper can purchase are eligible for merchant listings, that Google may attempt to verify merchant listing data before showing it, recommends Product markup in the initial HTML, and warns that JavaScript-generated markup can make Shopping crawls less frequent and less reliable for fast-changing price and availability.
Google Search Central: Merchant listing structured data, Technical guidelines · 2026-10-27 · documented
Claim limitation: Google crawling guidance. Other systems may read pages differently.

Google says providing both structured data and a Merchant Center feed maximizes eligibility and helps it correctly understand and verify your data, and that some experiences combine both, for example using feed pricing in product snippets when it is missing from the markup.
Google Search Central: Introduction to Product structured data · 2026-10-27 · documented
Claim limitation: Page last updated 2025-12-10.

Ahrefs tracked 1,885 pages that added JSON-LD between August 2025 and March 2026 against matched controls and reported no major uplift in citations in AI Overviews, AI Mode or ChatGPT, with a small AI Overviews decline it could not definitively attribute to schema. It notes all pages were already heavily cited, all schema types were pooled, the window was 30 days, and that there are still good reasons to use schema such as rich results.
Ahrefs: We Tracked 1,885 Pages Adding Schema. AI Citations Barely Moved. · 2026-10-27 · vendor-claim
Claim limitation: Not specific to Product markup or product pages; measures citations, not shopping selection. Published 2026-05-11, modified 2026-06-09; read 2026-10-27.

OpenAI says ChatGPT shopping results draw on structured metadata from first-party and third-party providers, such as price and product description.
OpenAI Help Center: Shopping with ChatGPT Search · 2026-10-27 · documented
Claim limitation: Does not say whether ChatGPT reads JSON-LD on merchant pages, or how it weights any source.

Schema.org defines ProductGroup as a group of products that vary only in well-described ways such as size, color or material, not itself directly offered for sale, and MerchantReturnPolicy as information about return policies associated with an Organization, Product or Offer.
Schema.org: ProductGroup; MerchantReturnPolicy · 2026-10-27 · documented
Claim limitation: Vocabulary definitions, not consumer requirements.

Google says shopping results labeled Top recommendations are selected based on relevance to the search, ratings, price and product features.
Google Shopping Help: Understand how shopping results are generated · 2026-10-27 · documented
Claim limitation: Does not mention structured data as an input or give weights.

Limitations

Google's structured data pages were checked on 2026-10-27. Requirements change; re-check them before acting.
The validation rules come from Google's pages for its own rich results. We found no ChatGPT, Copilot or Perplexity document saying whether or how they read Product markup on merchant pages; that is not documented on the pages we read.
Our worked check is a short script encoding a subset of Google's stated rules plus a visible-text comparison. It is not the Rich Results Test, and we did not run the hypothetical page through Google's tools.
The example store, product, prices and URLs are hypothetical.
The Ahrefs study is an observational vendor study of pages already cited by AI, pooled across schema types. It cannot show what Product markup does for product pages not yet visible to AI systems.
We don't document which Product markup any particular Shopify theme or app emits. Check your own page source.

Sources

Google Search Central: AI features and your website · accessed 2026-10-27
Google Search Central: Introduction to Product structured data · accessed 2026-10-27
Google Search Central: Merchant listing (Product, Offer) structured data · accessed 2026-10-27
Google Search Central: General structured data guidelines · accessed 2026-10-27
Schema.org: ProductGroup · accessed 2026-10-27
Schema.org: MerchantReturnPolicy · accessed 2026-10-27
Google Shopping Help: Understand how shopping results are generated · accessed 2026-10-27
OpenAI Help Center: Shopping with ChatGPT Search · accessed 2026-10-27
Ahrefs: We Tracked 1,885 Pages Adding Schema. AI Citations Barely Moved. · accessed 2026-10-27

What do agents receive about your product?

Send one product URL. We email your free shelf report within 24 hours: what agents received, what is missing, and the fix.

Email my shelf

Checked 2026-10-27 · For AI agents: this article as data