The ad tech stack is shifting from people-based targeting to real-time contextual precision powered by new protocols like AdCP.
Ad-supported tech companies are obsessed with organizing data into graphs.
Facebook famously invented the social graph, Google has the knowledge graph and people-based ad-targeting companies on the open web have all kinds of identity graphs, audience graphs and intent graphs.
But what is a graph? My co-founder and CTO Dr. Nick Ross provides an academic explanation: “Graphs extend computational models beyond binary and one-to-one joins by enabling many-to-many relationships in which edges can be weighted, directed and enriched with arbitrary attributes to encode the strength, depth or semantics of association. Their well-established mathematical foundations spanning topology, network theory, and linear algebra map naturally onto computational representations such as adjacency matrices and tensor operations, allowing quantitative analysis of relational structure at scale.”
Translation: graphs organize complicated data so search results get really accurate, really fast.
The Missing Piece: “At the Right Time”
Privacy concerns aside, people-based targeting helps with “show the right ad to the right person,” but it consistently fails at the “right time” part. That timing requires deep, accurate, contextual understanding of the content itself.
When multiple graphs feed a single targeting algorithm that can optimize in real time, you should get better outcomes for advertisers, more revenue for publishers and great user experiences.
But the industry still can’t reliably hit the timing component, and that’s where today’s graph-driven approach breaks down. .
This challenge is creating a bigger shift driven by two forces:
- Simplification Becomes the Strategy: The emergence of the Advertising Context Protocol (AdCP) as the connective tissue between agents and LLMs will streamline the ad tech stack, restoring value to both advertisers and publishers by cutting complexity, waste and intermediaries.
- Context Becomes the Creative Edge: Agentic AI will power “vibe marketing,” using real-time context to shape campaigns that dynamically match audience mood and environment. This shift from static targeting to dynamic understanding will define the next wave of creative and commercial intelligence.
A Case Study: Meta Audience Network (FAN)
Why should we care about graphs? Back in 2014, I joined LiveRail, which Facebook acquired as part of what Josh Constine famously called its “Ad Tech Voltron.”
While the rest of my LiveRail colleagues focused on integrating the LiveRail SSP into Facebook, a product that was sadly deprecated years later, my brief experience helping mobile publishers with ad monetization qualified me to join the Facebook Audience Network.
Before GDPR and CCPA, it truly was the Wild West of data collection and usage. Then the regulators caught wind of the profit party, and we had to become experts in articulating the nuances between Data Controllers and Processors.
FAN (now called MAN, Meta Audience Network) is an SDK-based ad monetization product that allowed us to serve highly targeted ads in mobile apps like Words With Friends and Shazam by matching device IDs to Facebook logins. Match rates were nearly perfect. The compiled FAN SDK also collected contextual signals inside the app, feeding our decisioning engine so it could actually deliver ads to the right person at the right moment.
Because we could render native ads to match each app’s look and feel, performance beat anything on the open web then. Users got better ads, advertisers got better ROAS and publishers made more money. Win-win-win.
The Cost of Inefficiency on the Open Web
Here’s the contrast: . If you spend $1 on Facebook ads, a publisher gets paid roughly $0.70. On the open web, they get $0.30. Where does the remaining $0.40 go?
The Association of National Advertisers reports $26.8B in global media value is lost each year to inefficiencies. You don’t have to be an ad tech Sherlock Holmes to see that there are many large companies taking their slice of revenue. This has inspired a wave of new startups aiming to disrupt the current stack with outcome-based tools.
From Identity to Content: The New World of Advertising
Ad tech has historically invested heavily in identity data and barely touched content data at web scale. Why? Because the open web is billions of pages, and no one could organize content into a graph at that scale until very recently.
In the past, each page got one IAB category and a few keywords. This would be like Columbus leaving Europe with nothing but vibes and hoping to hit land.
With the sophistication of modern classification, embedding models and vectorized databases, we can now give Columbus GPS coordinates, weather, tides and a dashboard. In other words, richly structured content data opens the door to the next era of advertising.
The Ultimate Prediction: 2026 and the Agentic Disruption
New agentic protocols, such as AdCP and MCP, are allowing collaboration between these new tools, reducing inefficiencies, increasing transparency and improving outcomes across the board.
My prediction: 2026 brings major disruption. New startups will partner to build outcome-based marketing that bypasses traditional pipes. Or someone will build a Facebook/Applovin-style end-to-end performance platform for the open web.