Amazon is reportedly considering a database of virtually every product sold online, including things it doesn't actually stock.
The proposed project, called World Product Graph, was described in an internal planning document reviewed by Reuters. The ambition is simple enough: make Amazon the place you research a purchase, even when the product is sold somewhere else.
That could be genuinely useful. It could also give Amazon considerably more influence over what we choose to buy.
Why would Amazon want everyone's product information?
Imagine you're buying a vacuum cleaner. You check Amazon's listings, visit a manufacturer's website to decipher the model numbers, then search for reviews to work out which one is worth buying.
Amazon wants more of that research to happen in one place.
According to Reuters, the proposed database would combine product information supplied by brands with information collected from retailers' websites. It could help explain which models are newer, how variants differ and which alternatives are worth considering.
That is more ambitious than simply adding extra products to a catalogue. It's about connecting the information shoppers currently piece together across several websites.
And if Amazon becomes the most convenient place to compare everything, there's less reason to leave.
Some of this is already happening
Amazon's existing Shop Direct programme already displays products from other retailers. In March 2026, Amazon said it covered more than 100 million products from over 400,000 merchants.
Shoppers can visit the retailer's own website or, for eligible items, use Amazon's Buy for Me service to have an AI agent complete the purchase.
In November, Amazon launched a merchant portal to make it easier for eligible shops to connect their catalogues.
There's also Alexa for Shopping, formerly called Rufus. Amazon says its assistant can answer product questions, suggest alternatives and help with purchases.
World Product Graph, if it ever launches, could give that assistant a much broader understanding of what's available.
The bigger question is who chooses what you see
Online shopping traditionally means searching, comparing listings and making a decision yourself.
An AI assistant can shorten that process. You might ask for comfortable headphones under £150, explain that you commute by train and get a shortlist instead of pages of results.
But someone still decides which products make that shortlist.
An assistant's answers depend on the information it can access, how it weighs competing products and whether it includes retailers outside its own ecosystem.
A comprehensive database could improve recommendations. It could also make Amazon an even more powerful gatekeeper for product discovery, including purchases that happen elsewhere.
That doesn't prove its recommendations would be unfair. It makes transparency about how those recommendations are produced more important.
This is a proposal, not a launch announcement
Reuters says the internal document carried an aspirational date of October 2027, but couldn't establish whether or when the project would launch.
Amazon told Reuters the document was outdated and might not reflect current plans. It also disputed the document's concerns about the accuracy of its shopping assistant, saying its own figures showed roughly 97% accuracy globally.
So World Product Graph could change substantially, or never become a public product at all.
Still, Shop Direct shows that Amazon is already looking beyond the things on its own shelves.
The interesting competition in online shopping may soon be less about who sells a product and more about who helps you decide which product to buy.
Sources
Every fact in this article comes from one of these.