Product data for manufacturers and brands
How AI agents tell your product from a look-alike: GTIN, MPN and brand, product data and answers to buyers’ questions. What we measure, and what we don’t.
Identity confidence
Besides the product identity check’s score, the result states how sure an agent can be that it found the exact product: a valid GTIN is “strong”, an MPN with the brand and no GTIN is “medium”, the product name alone is “weak”. Confidence doesn’t affect the score.
Product question coverage (free test, no effect on score)
An agent comparing products asks questions like size, material or warranty. The free test works out the product type and counts only the questions expected for it:
- apparel: size / dimensions, material / ingredients, care, returns, delivery time, stock / variants
- footwear: size / dimensions, material / ingredients, care, returns, delivery time, stock / variants
- accessories: size / dimensions, material / ingredients, care, warranty, returns, delivery time, stock / variants
- electronics: size / dimensions, compatibility, warranty, returns, delivery time, stock / variants
- beauty and personal care: size / dimensions, material / ingredients, returns, delivery time, stock / variants
- home and living: size / dimensions, material / ingredients, care, warranty, returns, delivery time, stock / variants
- a general product: size / dimensions, material / ingredients, care, compatibility, warranty, returns, delivery time, stock / variants
Each question is looked up first in structured data, then in the visible page text, with a rule-based dictionary and no AI; for answers found in the text we show the excerpt, because a phrase may not fully answer the question.
Every product, in the panel
The panel catalog gives each product a readiness score on six axes: price, buyability, identity (GTIN, brand), stock, attributes and returns; an axis that can’t be measured leaves the denominator. Missing attributes become tasks; values are never invented.
Scored checks
These checks are part of the agent readiness score; each links to its own page with the fix and how to verify it.
Product identity (GTIN/MPN, brand, SKU)
SC-DATA-03 · weight 5 · Product data
Why it matters: Agents match the same product across marketplaces and stores by barcode; a product without identity drops out of comparisons or loses to a marketplace listing.
What passes: Brand and a valid GTIN (barcode) or MPN are defined; the SKU is there too.
How the free test measures it
We read product identity from structured data: a valid GTIN (correct check digit) or an MPN, the brand and the SKU. An identifier plus the brand passes; none fails; partial credit weighs the identifier 50%, the brand 35% and the SKU 15%. The result also states identity confidence (no effect on score): a valid GTIN is “strong”, an MPN with the brand and no GTIN is “medium”, the product name alone is “weak”.
Product structured data (schema.org)
SC-DATA-01 · weight 7 · Product data
Why it matters: Google, ChatGPT Shopping and Perplexity read the product name, price and identity from this data first.
What passes: Every product page carries Product data as JSON-LD; products with variants use ProductGroup + hasVariant.
How the free test measures it
We look for JSON-LD Product (or ProductGroup) on the product page. Only microdata or Open Graph product tags is a partial pass; none fails and is a critical blocker.
Price, currency, availability and URL
SC-DATA-02 · weight 6 · Product data
Why it matters: A product whose price and stock can’t be verified drops out of the agent’s price filter and shortlist.
What passes: The offer carries price, an ISO currency code, availability and the product URL together.
How the free test measures it
We count four fields in the product’s offer: price, a valid ISO 4217 currency, availability and the offer URL. All four pass, none fails; anything in between earns points in proportion.
Can variants (size, color) be addressed?
SC-DATA-06 · weight 2 · Product data
Why it matters: An agent can fulfill “size M in navy” only if the variant is defined with its own URL and price.
What passes: Every variant opens at its own URL; structured data uses ProductGroup + hasVariant and variesBy.
How the free test measures it
Variants described as ProductGroup + hasVariant in structured data pass; if there are at least two color variants, each color needs its own distinct image, otherwise it’s a partial pass (size-only variants aren’t checked for images). If the page has a size/color selector but no such data: options with their own URLs are a partial pass, options without fail. Products without a selector are not applicable.
Images and description
SC-DATA-04 · weight 2 · Product data
Why it matters: Agents show a photo and a short description in the shortlist; without them the product is presented less well.
What passes: JSON-LD carries a stable https image URL and an original description; og:image is set as well.
How the free test measures it
We read the product image (structured data or og:image) and the description in the product’s structured data. A description of at least 20 words or 120 characters is enough; 10-19 words is short; fewer doesn’t count as a description. An image plus an adequate description passes; an image with a short description is a mostly partial pass; only one of the two is a half partial pass; neither fails.
What we don’t measure
- Whether a GTIN is registered to your brand: we check that the number is valid (its check digit), not who owns it.
- Dealer locator and where-to-buy pages: whether an agent can get from your site to a seller with stock.
- How your products are listed on retailers’ and marketplaces’ sites.