What Structured Data Your Product Page Needs in 2026
The rules for product schema used to be pretty simple. Add Product, Offer, and maybe AggregateRating, and you were done. Google was happy. Your rich snippets looked nice. You moved on.
Those days are over. AI shopping agents are reading structured data differently than Google's crawler does, and they need a much fuller set of fields to actually surface your products confidently. Here's the complete 2026 list for any ecommerce store that wants to stay visible.
The Core Five
Product. Must include: name, description (minimum 150 chars), image (at least one, ideally multiple), brand (as a nested Brand object with name), sku, gtin, mpn (if applicable), category, weight, color, material, size. The old "just name + image" version won't cut it anymore.
Offer. Must include: price, priceCurrency, availability (InStock/OutOfStock/PreOrder/etc), priceValidUntil, itemCondition (NewCondition/UsedCondition), url (to the product page), hasMerchantReturnPolicy, shippingDetails.
AggregateRating if you have reviews. ratingValue, reviewCount, bestRating, worstRating. If you don't have reviews yet, consider adding them with a honest count (even if small).
Review array. Include your latest 3-5 reviews as nested Review objects with author, datePublished, reviewBody, and reviewRating. Agents use these for social proof.
BreadcrumbList. Category path from homepage to product. This is easy but a surprising number of stores don't have it.
The Fields Agents Actually Care About in 2026
Here's the thing most SEO guides haven't caught up on yet. These are the fields that are becoming hard differentiators in AI agent retrieval:
hasMerchantReturnPolicy. This is the biggest one. ChatGPT's shopping answers now strongly prefer products from stores that explicitly declare return policies in schema. 94% of stores I've scanned are missing this. If you add it, you get a meaningful visibility boost.
shippingDetails. Include shippingRate, shippingDestination, shippingLabel, and deliveryTime. Agents read this to answer "can I get this by Thursday" type queries. Missing = you're out of the consideration set.
additionalProperty array. This is a generic key-value bag for attributes that don't fit the standard schema. Material composition, certifications, specifications, size charts. Agents love this because they can use it to filter and compare.
itemCondition. Sounds obvious but a lot of stores omit it. Especially important for used/refurbished/open-box items where agents need to know before recommending.
offers.priceValidUntil. If you run sales, this tells agents when the price expires. Without it, agents can't confidently recommend your sale price because they don't know if it's still valid.
Common Mistakes
A few things I see over and over.
Inconsistency between schema and on-page content. Your schema says the product is $29 but the page shows $35. Google's crawler forgives this, AI agents don't. They downweight you significantly for inconsistency.
Wrong availability values. Use the exact ISO values: https://schema.org/InStock, https://schema.org/OutOfStock, etc. Not "available" or "yes" or your own strings.
Shallow brand object. "brand": "Nike" is not valid. It should be {"@type": "Brand", "name": "Nike"}. Most themes get this wrong out of the box.
Missing mainEntityOfPage. This is a subtle one. It tells crawlers which URL is the canonical location for the schema. Helps with deduplication when you have multiple URLs for the same product.
Nested schema that isn't actually nested. If you have a product with offers, the offers should be inside the product schema, not in a separate JSON-LD block. Agents parse the whole document and sometimes miss stuff that's structured weirdly.
How to Validate
Google's Rich Results Test is still the gold standard for validation. https://search.google.com/test/rich-results. Run it on 5-10 of your highest-traffic product pages and fix any errors.
For more advanced validation, the Schema.org validator at https://validator.schema.org/ is more strict and catches things Google misses. Worth running periodically.
For AI-specific validation, there's no good public tool yet. The best approach is to query ChatGPT directly with a product search and see if your store shows up. If not, you probably have gaps somewhere in this list.
The Implementation Path
For Shopify stores, the fastest path to a compliant product schema is:
- Export 5 product pages, view source, find the JSON-LD block.
- Paste into https://jsonld.com/json-ld-visualizer/ or similar to see the structure.
- Compare against the full field list above. Note what's missing.
- Fix in your theme's product.liquid template, or use an app that adds the missing fields.
- Re-test with Rich Results Test until clean.
For custom stores (headless Shopify, Next.js, whatever), you'd do the same thing but in your product component. Next.js in particular makes this easy with the tag outputting JSON-LD.</p> <h2> <a name="the-takeaway" href="#the-takeaway" class="anchor"> </a> The Takeaway </h2> <p>Product schema in 2026 is richer and stricter than it was in 2022. The baseline for "visible to AI agents" has moved up significantly and most stores haven't caught up. If you invest a few hours updating your schema now, you'll be ahead of 80% of your competitors for a year or two. After that, it'll be table stakes and you'll wish you'd done it sooner.</p> <p>Run the Rich Results Test on your store this week. You'll probably find gaps. Fix them. It's one of the highest-ROI things you can do for your store's visibility in both google and AI channels right now.</p>