Attribution

Value creation happens at the edge of networks, but is not recognized, and is absorbed by platform companies that drive centralized networks. Despite the Internet being an open network powered by a common tongue that everyone can speak in - HTTP - it is still not possible to access your own data across various centralized networks such as Facebook or Google.

Attribution (marketing) - Wikipedia

Walled Gardens

HTTP networks on the Internet are able to block content creators (and their own users!) from accessing data about how their content is consumed by other users on the networks.

As each centralized network does this, it becomes impossible for businesses to get a single view of their network activity, as instead of a single network that one can track across, there are as many networks as there are platforms on the Internet, and more.

The Attribution Problem

The problem with attribution models is that in the best case, they are incomplete and therefore generally useless. Worst case, they're outright wrong and downright harmful.

The Bazaarvoice + Digiday Report: The Attribution Problem covers this in detail — the short version is that incomplete attribution is not just unhelpful but actively harmful, causing businesses to misallocate marketing spend toward the platforms that claim the credit rather than the people who created it.

Because it is not possible to find out which user on Facebook (say) shared your content with a potential shopper (just that the shopper happened to click on the CTA), businesses are forced to give the attribution to the platform as a whole (Facebook itself, and not Joe Blow the Facebook user).

And as a result, marketing has no choice but to pour more and more money into buying clicks from Facebook (and others), which convert at less than 1% typically, paying for a service that delivers less than 1% efficiency is likely to be labeled insane in any other industry.

But in online advertising: TINA. There Is No Alternative, not until the walls of the walled gardens on the Internet are brought down, or some other magic technology comes along that solves this very basic problem of the Internet.

Multi-Touch Attribution

If a business can figure out exactly who the people were that drove the customer to a purchase, they might be able to compensate them directly for the sale using available commission.

The problem, as described above is one of data. On the Internet, as a potential customer discovers content about your brand, we want to track not only where they engaged with the content, but also how they discovered it, and ideally even who shared it with them in the first place.

If we can create a network of such folks who share content with potential shoppers on the Internet, we can then directly pay them for any sales they generate. This might seem like the traditional affiliate model but is not.

Last-Click Attribution

Affiliate Networks rely on the last click which indicates the purchase in order to make payouts. This link-tracking and click-tracking depends on the use of 3rd-party cookies because affiliate networks are not first party but third parties.

If these affiliate networks had more integration across the Internet platforms, they would be able to track better, but the platform companies have no incentive to allow this type of access. They would be able to attribute value to not just the final click that drives a purchase at the end of a customer journey, but would be able to track all hops of that journey across the social Internet, and really create a data-driven machine that drive their business.

The Post-Cookie Era

Apple eliminated third-party app tracking starting with iOS 14.6, and the broader industry has since moved away from third-party cookie tracking across the open web. In the name of privacy, platforms tightened their grip on user data — most businesses, large and small, are now running partially blind on their marketing campaigns.

This shift has forced businesses to think harder about first-party networks and direct relationships. Brands that built their own communities and distribution are far less exposed; those that depended entirely on rented audiences and third-party tracking are left exposed by the change.

House of Cards

In many ways, today's Internet economy is a house of cards, relying on a single technical detail: Even though networks on the Internet are interoperable and free to use (thanks to the open HTTP protocol), these platforms block their own content creators from accessing information about the consumption of the content they create.

If this one technical challenge is overcome so that HTTP content can be meaningfully tracked across other HTTP networks, it would be a total game-changer for all existing business models on the Internet.

Deterministic Attribution for Agent Markets

The attribution problem gets harder as AI agents enter the picture. An agent that discovers a product, researches it, and recommends it across multiple sessions creates a chain of influence that is invisible to any single platform's tracking. Deterministic attribution — knowing exactly who and what drove a sale — becomes the foundation for any economy built on agents.

Intergraph is the institutional substrate for agent markets: a protocol layer that lets AI agents discover each other, exchange value, and coordinate across organizations. One of its core properties is deterministic attribution — the ability to trace value creation back to the agents and people who generated it, regardless of which platform or organization they belong to. This makes it possible to build coseller networks, agent commission structures, and community commerce systems that actually work at scale.