Design Product Syndication

Problem Statement: Product Syndication Platform

A product syndication platform takes a brand's catalog and publishes it, correctly shaped, to every marketplace the brand sells on — Amazon Seller Central, eBay, Google Merchant Center, Meta catalogs, and retailer-specific feeds. ChannelAdvisor, Salsify, and Feedonomics solve this; the core is fan-out with per-channel translation.

Golden record vs channel projections

The single most important distinction in the design: separate the authoritative catalog state from the channel-specific projections. The brand maintains one golden product record in a PIM (product information management) hub — the canonical truth of what the product is. Syndication then derives, per channel, a projection of that record that obeys each marketplace's category taxonomy, required attributes, and media rules. Amazon wants a bullet_point array and a browse-node id; Google wants a google_product_category and a GTIN; eBay wants an item-specifics map. The golden record holds the superset; each projection is a lossy, reshaped view of it.

Why the split is the spine

The reason this separation is load-bearing: the golden record changes for brand reasons (a new photo, a price change, a corrected description), while a projection changes for channel reasons (Amazon updated its required-attributes schema, or a category was remapped). Coupling them — editing Amazon-shaped data directly — means a brand edit has to know about every channel's quirks, and a channel change has to touch the source of truth. Keeping the golden record authoritative and the projections derived means brand edits flow outward to all channels automatically, and a channel's quirks stay contained in that channel's transform. This is the same authoritative-truth-plus-derived-views pattern the category uses everywhere, applied to multi-marketplace publishing.