Why it matters
Picture a distributor whose full catalog exists only inside SAP Ariba punchout sessions. ChatGPT cannot see a single SKU. A buyer researching a hydraulic fitting or an MRO part on ChatGPT never finds that distributor, because the part numbers live behind a procurement login. The punchout is great for the buyers already provisioned in Ariba or Coupa. It does nothing for everyone discovering suppliers through an AI engine first.
We call this a dark catalog. The data is rich and well-structured, and it is invisible to search and AI because it only renders after a cXML handshake. A crawler hits a punchout setup URL and gets a session redirect, not products. So the catalog never enters the index that AI engines retrieve from.
Punchout vs public catalog
A punchout catalog serves provisioned buyers inside their procurement system. A public, crawlable catalog serves discovery. You need both. The punchout closes the order for an existing account, and a crawlable catalog feeds the same part data, descriptions, and specs to an indexable page an AI engine can cite when a new buyer asks.
In practice
A hydraulics distributor running a cXML punchout into Coupa exports the same PIM records to public product pages, one URL per SKU with the part number, spec table, and cross-references in the markup. The punchout keeps serving its provisioned accounts. The public catalog gives ChatGPT and Perplexity something to retrieve, so a buyer who starts at an AI engine can reach the same parts.