Reuters reported on Friday 13th November 2026 that Amazon has an internal plan called World Product Graph.

It would collect data on virtually every item for sale, whether Amazon stocks it or not.

Amazon says the document is a draft, outdated brainstorming, and Reuters says plans like this can be changed or dropped. So treat it as direction, not roadmap.

The part that caught my eye

The article says Coca-Cola and Bayer have signed up to contribute product data, and P&G was involved in a trial. Amazon would also crawl other retailers’ sites. No one knows though on whether permission would always be sought or they just crawl.

Those same names are SitecoreAI DAM (Content Hub) enterprise customers.

For me that matters massively.

Brands like these have spent years building a governed source of truth in a data model made of entities and the relations between them: assets, product content, approvals.

Amazon is trying to rebuild that picture from the outside. The document itself says the graph depends on discovering how brands organise their products on their own websites.

Sitecore literally has that map already.

Crawled or contributed

This is the real choice for a brand. You can let a platform piece your product story together from page scrapes, or you can offer it a curated, approved feed. The second route puts the brand in control of what is said and which version of a product is current.

Amazon says it wants accuracy, and the document admits its own assistant gave wrong or incomplete answers. A clean feed from the system of record helps with that.

Where MarketingIQ fits

I wrote last week about Sitecore MarketingIQ, the intelligence layer shown at Symposium. Its two jobs are holding brand intelligence and orchestrating marketing work, and it reaches outside tools through APIs and MCP.

Today that is mostly about getting marketers’ requests in. My question is whether it can work in the other direction too.

If brand context, approved product knowledge and asset relationships already sit in SitecoreAI DAM, an outbound interface could expose them in a governed way.

That could serve Amazon’s graph, but also Meta’s Muse, Google, or whichever agent a shopper uses next. Intelligence on the way in, visibility on the way out.

My early call

This is my call out, and I’m putting it on the record now.

Sitecore, if this is a product you’d want to explore, I’m happy to talk. Give me a call.

What I’d do on a project now

  • Audit what your DAM and content platforms already know. Product relationships, generations and approved claims are often richer than the website shows.
  • Decide your terms before someone crawls you. What you share, under what licence, with what attribution.
  • Expose it properly. An API or MCP layer with clear versioning beats hoping a crawler infers it.
  • Govern the outbound side. Every external feed needs an owner and an approval trail.
  • Measure beyond visits. Fewer visits does not mean less influence.

My take

Feeding the platform that may sit between you and your customer is a strategic decision, not a technical one.

And I’m cautious on the vendor side too: a good set of vision slides is not a good rollout.

But the brands best placed for agent-led shopping will be the ones that already treat their content as governed data.

Some of them have been doing it for years

Leave a Reply