Which schema markup types matter most for getting products cited by ChatGPT and Perplexity?

In our extraction testing (n=200 pages per schema type, directional), Product schema landed a 39% overall citation rate, with a meaningful engine split: 49% on Perplexity versus 38% on ChatGPT and 31% on Claude. SoftwareApplication performed better at 44% overall (52% on Perplexity) for tool and app products. But the highest-performing types on product SITES were not product types at all: FAQPage (71%) and HowTo (66%) on collection and guide pages outperformed Product schema on product detail pages. The winning combination is Product with nested Offer and AggregateRating on detail pages, plus FAQPage answering buying questions on collection pages.

What fields does Product schema need for AI citation?

The completeness bar is higher than Google's rich-result minimum. Include name, description, image, brand, and sku on the Product, a nested Offer with price, priceCurrency, and availability, and a nested AggregateRating with ratingValue and reviewCount when you have genuine reviews. Incomplete Product schema (name and image only, which is common on default storefront themes) rarely gets extracted: the engine has a product entity but nothing decision-relevant to cite it for. Price, availability, and rating are the fields AI shopping answers are actually built from.

Does Review schema help AI citations?

By itself, marginally: standalone Review schema scored 30% in our tests and is often ignored unless tied to an Article or Product entity. Nested inside Product as AggregateRating plus individual Review entries, it contributes decision-relevant data (rating, review count) that AI shopping answers cite. The practical rule: never ship Review as a standalone block on a product page; nest it in the Product entity.

Why does Perplexity cite product pages more than ChatGPT?

Perplexity's query mix skews toward product and tool research, and its extraction weights Product and SoftwareApplication schema visibly higher than ChatGPT or Claude in our per-engine data (Product: 49% Perplexity versus 38% ChatGPT versus 31% Claude). If you sell products, Perplexity is the engine where schema investment pays off fastest, and it is also the engine most likely to cite a smaller store when the structured data is complete, because it retrieves per query instead of leaning on training-data brand recognition.

Do Shopify stores need extra schema work?

Usually yes. Most Shopify themes ship a partial Product schema (name, image, sometimes price) and stop there: no availability, no AggregateRating wiring, no FAQPage on collection pages, and Organization schema missing entity fields. The gaps are fixable at the theme level. The three checks that matter for AI shopping visibility are Product/Offer/AggregateRating completeness on product pages, collection pages that answer 'best X for Y' questions in extractable form, and entity recognition signals (Organization schema, consistent business identity, third-party citations).

Which Schema Types Get Products Cited by ChatGPT and Perplexity?

Product schema alone will not get your products cited: it scored 39% in our extraction tests, well behind FAQPage at 71%. The stack that performs for product sites is two-layered: a complete Product entity (nested Offer with price and availability, nested AggregateRating) on every product detail page, plus FAQPage schema answering buying questions on collection pages. And the engine split matters more here than anywhere else in our data: Perplexity extracted Product schema on 49% of tested pages versus ChatGPT's 38% and Claude's 31%, which makes Perplexity the engine where product schema investment pays off first. This post is the product-specific depth on the schema types AI engines actually use, applied to the pages where purchases happen; the conceptual model behind why structure beats authority for citations is in the five-pillar framework for AI-citable sites.

What the Data Says About Product-Relevant Schema#

From our extraction testing (20 correctly-implemented pages per schema type per platform, n=200 per type; directional, not a controlled experiment; methodology in the full schema test):

Schema type	ChatGPT	Perplexity	Claude	Overall	Product-site takeaway
FAQPage	68%	71%	74%	71%	Your collection pages' weapon, not your product pages'
HowTo	63%	65%	69%	66%	Buying guides and setup content
SoftwareApplication	41%	52%	38%	44%	Use instead of Product for tools and apps
Product	38%	49%	31%	39%	Necessary on detail pages, insufficient alone
Review	29%	34%	27%	30%	Only pays when nested in Product

Independent research points the same direction at much larger scale: the AirOps analysis with Kevin Indig (353,799 pages) found JSON-LD pages cited 38.5% of the time versus 32% without, and stresses the same caveat we do: schema correlates with citation because it reflects extractable, answer-first structure, it does not cause citation by itself.

Two readings for product sites. First, the product types are mid-table: complete them because AI shopping answers are assembled from their fields (price, availability, rating), not because the markup alone wins citations. Second, the highest-rate types (FAQPage, HowTo) are content types, which means a product site's biggest schema win is usually on its collection and guide pages, not its product detail pages.

The Product Entity That Gets Extracted#

The completeness bar is higher than Google's rich-result minimum. This is the shape that carries every field an AI shopping answer is built from:

{
  "@context": "https://schema.org",
  "@type": "Product",
  "name": "Trail Runner Pro 3",
  "description": "Lightweight trail running shoe with a rock plate and 6mm drop, for technical terrain.",
  "image": "https://example.com/products/trail-runner-pro-3.jpg",
  "brand": { "@type": "Brand", "name": "Example Running Co" },
  "sku": "TRP3-BLK-42",
  "offers": {
    "@type": "Offer",
    "price": "129.00",
    "priceCurrency": "USD",
    "availability": "https://schema.org/InStock",
    "url": "https://example.com/products/trail-runner-pro-3"
  },
  "aggregateRating": {
    "@type": "AggregateRating",
    "ratingValue": "4.6",
    "reviewCount": "212"
  }
}

The fields most commonly missing on storefront themes we scan are availability, aggregateRating, and brand. A Product block with only name and image gives the engine a product entity with nothing decision-relevant attached: it knows the product exists but has no price, stock state, or rating to recommend it with. That is the product-page version of being recognized but not cited.

Two placement rules from the extraction data. Review schema standalone scored 30% and is often ignored; the same review data nested as aggregateRating inside the Product entity is what shopping answers actually surface. And if your product is software or an app, use SoftwareApplication instead of Product: it outperformed Product in every engine, with the biggest gap on Perplexity (52%).

Collection Pages Are Where Product Sites Get Cited#

AI shopping queries are mostly comparative: "best [category] for [use case]", "alternatives to [competitor]", "is [product] worth it". Those queries do not resolve to a product detail page; they resolve to whichever page answers the comparison in extractable form. That is a collection or guide page with FAQPage markup, and at 71% extraction it is the highest-leverage schema move a product site can make.

The pattern: give each collection page 3 to 5 question-answer blocks that answer real buying questions in 50 to 150 words each, marked up as FAQPage, with the answer text matching the visible page content. A collection page that answers "best trail running shoes for rocky terrain" in an extractable block is competing for the query your product detail pages never could.

What Product Schema Cannot Fix#

The same infrastructure caveat as everywhere else in this cluster, sharpened for storefronts: schema that is rendered client-side is schema that does not exist. JavaScript-heavy storefront themes and headless setups that inject JSON-LD after hydration serve AI crawlers an empty block. Verify with curl, not with a browser: what curl sees is what the crawler sees, and what AI crawlers see is frequently not what your team sees. Fix rendering first, then complete the schema stack.

Check Your Store's Stack#

If you run a Shopify or other product store, the AI Shopping Visibility scan tests exactly this surface: Product/Offer/Review/AggregateRating completeness on product pages, collection-page answerability for the "best X for Y" queries in your category, and entity recognition signals, with a head-to-head competitor citation comparison. For any other site type, the free scan covers the 10 structural signals including structured data completeness. For the deeper vocabulary mapping of which Schema.org types each answer engine prefers, see Schema.org for Answer Engines on chudi.dev, and for the category context, Answer Engine Optimization Explained.