Ecommerce Structured Data - Product, Offer, Variants (ProductGroup) and BreadcrumbList in Practice
RETURN_TO_BLOG
AI & SEO 16 min

Ecommerce Structured Data - Product, Offer, Variants (ProductGroup) and BreadcrumbList in Practice

Paweł Wiszniewski
Paweł Wiszniewski
SEO & GEO Specialist · AI Engineer

Structured data in an online store tells Google what you sell, at what price, whether it is available and how shipping works. The rule: every product page that can be bought needs Product with an Offer (a price above zero and a currency) in the initial HTML. Always add a GTIN when the product has an EAN code, and describe return and shipping policies once for the whole store (Organization or Merchant Center), not on every product page. Use ProductGroup only when a product has at least two variants with their own SKU, price or availability. The most common mistake I found in Polish stores was not missing markup but a price in the data that differs from the one the shopper sees.

What structured data an online store needs: merchant listing vs product snippet, required Product and Offer properties, returns and shipping in Organization, variants with ProductGroup, breadcrumbs, ratings, JSON-LD vs JavaScript, an audit of 36 pages in 19 Polish stores, a script and the sale-price pitfall.

/// THIS POST IN SHORT

Ecommerce structured data - Product, Offer, variants

This post covers how merchant listings differ from product snippets, which Product and Offer properties are required and which are worth adding, where to describe returns and shipping, how to mark up variants with ProductGroup, breadcrumbs and ratings, whether data can be injected with JavaScript, the results of my audit of 36 product pages in 19 Polish online stores, a script for checking the data and the sale-price pitfall.

What structured data does for a store

Product data lets Google show a product in results with its price, availability, rating and shipping information, as well as in the Google Shopping knowledge panel, Google Images and popular products results. This is how Google illustrates it in its documentation:

An example product panel with prices from several stores - screenshot of Google Search Central documentation (merchant listing structured data), Nov 2026
An example product panel with prices from several stores - screenshot of Google Search Central documentation (merchant listing structured data), Nov 2026

Structured data is not a ranking factor that lifts a page's position. It produces a result that tells the shopper more, and in product results and AI shopping it decides whether your offer can appear at all. I cover Schema.org markup in general in the structured data guide.

Do you need structured data if you have Merchant Center - the first decision

The most common question. Google's answer: ideally both. A Merchant Center feed and on-page data complement each other rather than replace each other.

OptionWhat it gives youWhat it will not doWhen it is enough alone
Merchant Center feedThe Shopping tab, free product listings, ads; Google can take the price from it when the page has noneIt does not describe the page for other search engines and AI assistants that read the page codeNever fully - Google recommends both
Product structured data on the pageRich results in Search, feed verification against the page, data for other crawlersIt will not get products into the Shopping tab - that needs Merchant CenterA store that does not use the Shopping tab or product ads
BothThe best eligibility for every result format-Recommended for every store

Google says plainly that providing both on-page structured data and a Merchant Center feed maximizes eligibility. I cover Merchant Center setup in the post on ecommerce SEO and GEO.

Google documentation on Product data: structured data on the page plus a Merchant Center feed maximizes eligibility - screenshot of Google Search Central, Nov 2026
Google documentation on Product data: structured data on the page plus a Merchant Center feed maximizes eligibility - screenshot of Google Search Central, Nov 2026

Product snippet or merchant listing

Google distinguishes two kinds of product results, and that decides which properties are required.

FormatFor which pagesRequiredNotes
Merchant listingPages where the product can be boughtname, image, offers with price (above zero) and priceCurrencyThis is the format for a store
Product snippetPages where the product cannot be bought: reviews, comparisons, rankingsname and one of: review, aggregateRating, offersThe format for blogs and review sites

A store product page that meets the merchant listing requirements is also eligible for product snippets, so a store does not need to add anything separately.

Product, Offer and Organization - which properties are required

Product, Offer and Organization properties: required for merchant listings and recommended
Product, Offer and Organization properties: required for merchant listings and recommended
PropertyWhereRequiredPractical note
nameProductYesThe same name as the H1, without additions like "sale"
imageProductYesA product photo, not a logo or banner
offers.priceOfferYes (above 0)A number with a dot: 149.99, without a currency sign or spaces
offers.priceCurrencyOfferYesPLN, EUR etc. in ISO 4217 format
offers.availabilityOfferRecommendedInStock, OutOfStock, PreOrder, BackOrder, Discontinued and others; one value
itemConditionOfferRecommendedNewCondition for new products
gtin / gtin13ProductRecommendedThe EAN code from the packaging; the most specific one you have
brand.nameProductRecommendedThe manufacturer's brand, not the store name
sku, mpnProductRecommendedThe store's own code and the manufacturer's code
aggregateRating, reviewProductRecommendedOnly genuine reviews visible on the page
shippingDetails, hasMerchantReturnPolicyOffer or OrganizationRecommendedIdeally once in Organization, see below

Returns and shipping: on the product page or once for the store

Google recommends describing the standard return and shipping policies once, in Organization markup (MerchantReturnPolicy and ShippingService), and only exceptions for specific products at Offer level. When the same information is in several places, Google uses it in this order, from strongest:

  1. 1.Content API for Shopping,
  2. 2.Merchant Center or Search Console settings,
  3. 3.product-level markup (Offer),
  4. 4.Organization-level markup.

The practical takeaway: if your return and shipping policies are set in Merchant Center or Search Console, they win and you do not need to repeat them on every product page. If not, add them once in Organization on the home page - a few lines instead of thousands of repetitions.

Product variants: ProductGroup

Sizes and colours of one model are described with ProductGroup: the group has a productGroupID (the parent product code) and variesBy (how the variants differ), and hasVariant holds the individual products, each with its own SKU or GTIN, price and availability.

ProductGroup and variants: one group, each variant with its own SKU, GTIN, price and availability
ProductGroup and variants: one group, each variant with its own SKU, GTIN, price and availability
SetupWhenGoogle's requirements
One page for all variantsSize and colour chosen on one product pageOne canonical URL for the group; each variant can be preselected with a URL parameter, e.g. ?color=black&size=38
A separate page per variantColours with demand and their own URLEvery page has full markup and repeats the ProductGroup definition
No ProductGroupA product without variants, or variants that do not differ in price or availabilityA plain Product with Offer

An example for one page with variants (shortened to two variants):

product-group.json
{  "@context": "https://schema.org",  "@type": "ProductGroup",  "name": "Trek 12 hiking boots",  "url": "https://yourstore.com/trek-12-hiking-boots",  "brand": { "@type": "Brand", "name": "Trek" },  "productGroupID": "BT12",  "variesBy": ["https://schema.org/color", "https://schema.org/size"],  "hasVariant": [    {      "@type": "Product",      "name": "Trek 12 hiking boots black 38",      "sku": "BT12-BK-38",      "gtin13": "5901234123457",      "color": "black",      "size": "38",      "image": "https://yourstore.com/img/bt12-black.jpg",      "offers": {        "@type": "Offer",        "url": "https://yourstore.com/trek-12-hiking-boots?color=black&size=38",        "price": 149.00,        "priceCurrency": "PLN",        "availability": "https://schema.org/InStock",        "itemCondition": "https://schema.org/NewCondition"      }    },    {      "@type": "Product",      "name": "Trek 12 hiking boots black 39",      "sku": "BT12-BK-39",      "color": "black",      "size": "39",      "image": "https://yourstore.com/img/bt12-black.jpg",      "offers": {        "@type": "Offer",        "url": "https://yourstore.com/trek-12-hiking-boots?color=black&size=39",        "price": 149.00,        "priceCurrency": "PLN",        "availability": "https://schema.org/OutOfStock"      }    }  ]}

The GTIN and SKU codes in the example are illustrative. In a real store every variant must have a unique identifier - Google requires it explicitly. I cover when a variant should get its own indexable URL in the post on ecommerce indexing.

Google requirement: each variant must have a unique ID in its structured data (for example sku or gtin) - screenshot of Google Search Central documentation, Nov 2026
Google requirement: each variant must have a unique ID in its structured data (for example sku or gtin) - screenshot of Google Search Central documentation, Nov 2026

Breadcrumbs: a BreadcrumbList must have at least two ListItem elements, each with position and name, and all except the last also with item (a URL). Google shows this feature on desktop. Breadcrumbs in the data should match the breadcrumbs visible on the page and the category path the product really belongs to.

Ratings and reviews: the self-serving review restriction applies to LocalBusiness and Organization markup - a business will not get stars for reviews about itself on its own site. Product reviews on a store's product page are allowed, as long as they come from buyers, are visible on the page and are neither fake nor undisclosed paid reviews.

JSON-LD or microdata, in the HTML or via JavaScript

MethodCan Google read itRecommendation
JSON-LD in the initial HTMLYes, immediatelyRecommended - the easiest to maintain
Microdata in the HTMLYesWorks, but harder to keep consistent through template changes
JSON-LD added by JavaScriptYes, after renderingGoogle warns that for Product this can make Shopping crawls less frequent and less reliable
JSON-LD from Google Tag ManagerYes, after renderingA risk of mismatch with the page; Google advises caution

For a store there is one conclusion: Product in the initial HTML. An extra argument: many AI assistant crawlers do not run JavaScript, so they will not see data added by a script at all.

Google warning: dynamically generated Product markup can make Shopping crawls less frequent and less reliable - screenshot of Google Search Central documentation, Nov 2026
Google warning: dynamically generated Product markup can make Shopping crawls less frequent and less reliable - screenshot of Google Search Central documentation, Nov 2026

What my audit of 19 Polish stores showed

How I checked: in November 2026 a script sampled product URLs from the sitemaps of 23 Polish online stores (electronics, drugstores, pharmacies, fashion, books, home and garden, pet supplies) and opened them in a Chromium browser. It counted a page as a product page if it had an add-to-cart button - regardless of markup. It took two such pages per store. For each it read the JSON-LD after rendering and in the raw HTML, checked the GTIN check digit and compared the price in the data with the prices visible on the page. I have complete results for 36 pages in 19 stores.

Audit results: the share of 19 Polish stores providing each type of product data
Audit results: the share of 19 Polish stores providing each type of product data
What I checkedResultNotes
Product data in JSON-LD17 of 19 stores2 stores have no Product in JSON-LD; one has microdata only
Product already in the raw HTML32 of 32 pages with dataNo store added Product only with JavaScript
GTIN code9 of 19 storesAll 18 codes found had a valid check digit
Item condition (itemCondition)12 of 19-
Shipping cost in the data8 of 19At Offer level; only 2 pages had shipping in Organization
Return policy in the data7 of 19Some stores may set it in Merchant Center - I did not check that
Product rating (aggregateRating)5 of 19Some stores show reviews without markup
BreadcrumbList14 of 19-
Variants via ProductGroup4 of 19jysk.pl, agatameble.pl, doz.pl, zooplus.pl
Price in the data matches the page28 of 31 pages3 mismatches described in the pitfall below

This is what correct data looks like in the Schema.org validator for one of the audited pages - Product with GTIN, SKU and item condition in the raw HTML:

The Schema.org validator for a bee.pl product page: WebPage with Product (gtin13, sku, itemCondition) in the page code - screenshot, Nov 2026
The Schema.org validator for a bee.pl product page: WebPage with Product (gtin13, sku, itemCondition) in the page code - screenshot, Nov 2026

The takeaway: almost every large store has the basics (Product, price, image, brand). The differences are in the data that helps win a comparison: GTIN, shipping and returns appear in half of the stores or fewer, and only 4 stores of 19 describe variants with ProductGroup. Limitations: two pages per store, and return and shipping policies may be set in Merchant Center, which I cannot see.

How much work - two scales

Estimates from my quotes (November 2026) for implementing the data in store templates, excluding preparing source data (GTINs, photos).

AssumptionStore A: 500 products, no variantsStore B: 20,000 products with variants
Product and Offer in the product template3-6 hours6-12 hours
Returns and shipping in Organization1-2 hours2-4 hours (several shipping methods, exceptions)
ProductGroup for variantsnot applicable10-20 hours
BreadcrumbList1-2 hours2-4 hours
Testing on a sample and Search Console monitoring2-3 hours6-10 hours
Totalabout 7-13 hoursabout 26-50 hours

On platforms such as Shopify or WooCommerce, the theme or a plugin adds some of the data, but usually without GTIN, shipping, returns and variants. I describe the typical gaps in WooCommerce on the WooCommerce SEO page, and I implement full data in store templates as part of my ecommerce SEO work.

Script: check product data on a sample of pages

The script takes a list of product page URLs, extracts Product from the JSON-LD in the HTML and checks: the required properties, a price above zero, the currency, availability, the GTIN check digit and whether the price in the data appears among the prices shown in the page text.

check_product.py
# pip install requestsimport json, re, sysimport requestsdef blocks(html):    for m in re.finditer(r'<script[^>]+application/ld\+json[^>]*>(.*?)</script>', html, re.S | re.I):        try:            data = json.loads(m.group(1))        except ValueError:            yield {"_error": "invalid JSON"}; continue        stack = [data]        while stack:            x = stack.pop()            if isinstance(x, list): stack.extend(x)            elif isinstance(x, dict):                yield x                stack.extend(v for k, v in x.items() if k in ("@graph", "mainEntity") and isinstance(v, (list, dict)))def gtin_ok(code):    d = re.sub(r"\D", "", str(code))    if len(d) not in (8, 12, 13, 14): return False    s = sum(int(c) * (3 if i % 2 == 0 else 1) for i, c in enumerate(reversed(d[:-1])))    return (10 - s % 10) % 10 == int(d[-1])for url in [u.strip() for u in open(sys.argv[1]) if u.strip()]:    html = requests.get(url, headers={"User-Agent": "Mozilla/5.0"}, timeout=20).text    text = re.sub(r"<[^>]+>", " ", html)    prods = [b for b in blocks(html) if b.get("@type") in ("Product", "ProductGroup")]    if not prods:        print("NO Product in HTML:", url); continue    p = prods[0]    if p.get("@type") == "ProductGroup" and p.get("hasVariant"):        p = p["hasVariant"][0]    offer = p.get("offers") or {}    offer = offer[0] if isinstance(offer, list) else offer    problems = [f for f in ("name", "image") if not p.get(f)]    price = offer.get("price")    if not price or float(price) <= 0: problems.append("price missing or 0")    if not offer.get("priceCurrency"): problems.append("no currency")    if not offer.get("availability"): problems.append("no availability")    gtin = next((p[k] for k in ("gtin13", "gtin", "gtin14", "gtin12", "gtin8") if p.get(k)), None)    if gtin and not gtin_ok(gtin): problems.append(f"bad GTIN check digit {gtin}")    if price:        shown = {f"{float(price):.2f}".replace(".", ","), f"{float(price):.2f}"}        if not any(s in text for s in shown): problems.append(f"price {price} not found in page text")    print("OK" if not problems else "PROBLEMS", url, "; ".join(problems))

Note: the script reads raw HTML, so prices inserted with JavaScript can trigger a false alarm in the price comparison. In that case check the page by hand or use a headless browser (e.g. Playwright), as in my audit.

Production pitfall: a price in the data that differs from the page

In 3 of 31 audited pages the price in the data did not match the price the shopper saw. In all three cases the data and the visible price apparently came from different places in the store.

StorePrice in the dataPrice on the pageLikely cause
zooplus.plPLN 20.77PLN 25.96 (selected variant)The data gives the price of a different variant than the one selected on the page
ezebra.plPLN 89.00PLN 69.99The data has the regular price, the page shows the sale price
doz.plPLN 51.99PLN 59.79The data gives a different price than the page; I could not establish why from outside

Google says it may verify product data before showing it in results, and for data added through Tag Manager it warns about the risk of a mismatch between page content and markup. A price mismatch is a quick route to disapproved offers in Merchant Center and to a result that promises a different price than the cart.

The fix: the price in JSON-LD must come from the same source and the same logic as the displayed price - sale price included. With variants, the Offer must refer to the variant selected at that URL. Add a test in the template or in monitoring (the script above) that compares both prices on a sample of pages after every price list change.

What not to do

  • Do not put a currency sign or spaces in price - only a number with a dot, with the currency separately in priceCurrency.
  • Do not use a price of 0 for unavailable products - use availability: OutOfStock.
  • Do not add the data only through Tag Manager or a script after page load.
  • Do not repeat return and shipping policies on thousands of pages if you can provide them once in Organization or Merchant Center.
  • Do not mark up ratings that are not on the page, and do not make up reviews.
  • Do not give all variants the same SKU - every variant needs a unique identifier.
  • Do not add GTINs without checking the check digit.

FAQ

I describe the order of all store work in the ecommerce SEO strategy. If you want me to review the structured data in your store and implement fixes in its templates, see ecommerce SEO.

/// AUTHOR
Paweł Wiszniewski – AI & Web Engineer

Paweł Wiszniewski

SEO & GEO Specialist & AI Engineer

SEO/GEO specialist (10 years) and AI engineer (3 years). I build search visibility, AI systems and automations that reduce costs and improve operational efficiency.

Signal received?


Book a free consultation

Initiate protocol. Establish connection. Let's build something loud.

> WAITING_FOR_INPUT...

Form data is controlled by E-Marketing Paweł Wiszniewski (Suwałki, Poland) and used only to reply and prepare an offer. Details and your rights: Privacy policy.