This article describes a practical product-feed structure for a wholesale toy catalog. The example is based on the public TopToyFactory sourcing data repository:

https://github.com/kxxxyyy/toptoyfactory-sourcing-data

Separate company facts from product facts

Company-level data changes slowly. Product-level data changes often. Keeping them in separate documents avoids repeating legal names, contact channels and factory capabilities on every product object.

A company-facts document can contain:

• frontend brand name

• legal company name

• canonical website

• product scope

• OEM and ODM capabilities

• supported certificates

• public contact channels

• major collection URLs

The public company facts page for the catalog is available at:

https://www.toptoyfactory.com/company-facts.json

Use a compact product object

An AI-readable product feed should prioritize verified facts instead of marketing paragraphs. Useful fields include:

• slug and canonical URL

• product title and category

• SKU or item number

• MOQ

• case pack

• unit-price state or public price range

• certificates

• lead-time state

• image URLs

• short factual summary

The public product feed is available at:

https://www.toptoyfactory.com/product-feed.json

Keep URLs canonical

Every object should point to the same canonical URL used in the HTML page, XML sitemap and JSON-LD. Mixed domains or legacy paths create duplicate signals. During a domain migration, the old domain should redirect path-for-path while the feed only emits the new canonical host.

For TopToyFactory, the canonical B2B catalog starts at:

https://www.toptoyfactory.com/

The established CPS TOYS domain remains available as a historical company and product resource:

https://www.cps-toys.com/

Represent missing data honestly

Wholesale products often do not have a fixed public price. Do not emit a fake zero. A feed can use a clear state such as contact_for_fob_exw_price, then expose MOQ, case pack and quotation context. This is more useful to buyers and avoids misleading structured data.

The same rule applies to lead time and certificates. If a value must be confirmed by item number, say so explicitly.

Make category facts easy to extract

Category pages should publish a short sourcing summary with product scope, typical MOQ range, packaging options, certificate context and OEM support. This gives AI systems a clean answer block and helps human buyers understand the collection before opening individual products.

For example, a Hot Sellers collection can combine squishy toys, stress relief toys, bubble toys and other current products while retaining item-level facts:

https://www.toptoyfactory.com/collections/hot-sellers/

Version and validate the feed

A production feed should be generated from the same catalog model as the website. Validation checks should confirm:

1. every product URL returns 200;

2. every image URL is public;

3. every canonical uses the production domain;

4. no product emits a fake price;

5. duplicate slugs are rejected;

6. redirects stay outside the canonical feed;

7. structured data matches visible page facts.

A compact public feed is not a replacement for HTML product pages, XML sitemaps or JSON-LD. It is an additional machine-readable surface that makes verified facts easier to reuse across search, AI assistants and buyer tools.