AI Cannot Identify My Shopify Product: A Field-Level Diagnostic

Start with the failed identity

Write the exact wrong result:

Expected: Axiforma font family with style packages and license choices
Observed: generic downloadable product

Without that distinction, “improve product description” is not actionable.

Three public identity failures

MyFonts: product semantics absent from initial JSON-LD

The Axiforma page visibly exposed family name, designer, publisher, 20 styles, individual-style price, family-package price, and licensing choices. Our initial response contained three JSON-LD blocks but no detected Product entity.

The page is rich for a person, but a conventional physical-product parser may not recover the digital offer model.

Gymshark: visible apparel facts without initial Product entity

The Train T-Shirt page exposed product name, black colour, regular fit, XS–3XL sizes, price, and review-derived fit guidance. Our initial JSON-LD inspection did not detect a Product entity.

A system may still render or infer the page, but the server-response path has weaker explicit product semantics.

HexClad: Product exists with a polluted name

The six-piece pot set had a Product entity, SKU, GBP offer, and availability. Its name ended with Default Title, an internal Shopify variant label.

Identity can fail through absence or contamination.

Diagnostic sequence

url='https://example.com/products/example-product'
curl -sSL "$url" -o product.html
rg -n '<title|<h1|rel="canonical"|application/ld\+json|ProductGroup|"Product"|sku|category|brand' product.html

Record:

Layer	Expected evidence
URL	stable canonical product path
title/H1	specific public product name
first paragraph	what the product is and who it is for
specifications	discriminating attributes
Product graph	identity and offers
collections	useful parent category, not campaign only
Shopify taxonomy	standard product category where applicable

Check Shopify catalog fields

In Shopify admin:

Products → select product

Review:

• title
• product category
• product type
• vendor
• options and variants
• SKU and barcode
• collections
• typed metafields

Do not place skill level in vendor, promotional collection names in category, or a family-level SKU on every size variant.

Build an identity statement

Use a factual sentence:

Axiforma is a 20-style font family published by Monotype,
available as individual styles or a complete family with separately selected licenses.

For a Shopify physical product:

Battery Video Doorbell is a rechargeable Australian-market doorbell
for 2.4 GHz Wi-Fi, with optional 8–24 VAC continuous charging.

The sentence should expose category and differentiating constraints without keyword repetition.

Digital and bundle products need different models

MyFonts offers vary by style package and license rights, not colour and size. HexClad’s bundle has a parent offer and included items. A classifier that assumes every product has physical variants will produce bad competitors.

Allow an unknown or needs_review state when the commerce model is ambiguous.

Expected result

A public reviewer can recover:

• canonical name
• brand or publisher
• stable category
• offer type
• variant or package dimensions
• decisive compatibility or eligibility facts

Verification prompts

What is this product?
Which category should it be compared within?
Which facts distinguish its variants or packages?
What should not be inferred from the page?

Run those questions against the public page evidence before testing an external answer engine.