We Set Out to Solve Measurement. We Built Infrastructure.
We did not pivot from measurement to chase AI. We followed measurement to the fragmented, inconsistent, and incomplete data beneath it—and built the infrastructure required to solve it.
by Stephen Williams, CEO
Marketing Evolution was founded 25 years ago on a simple thesis: better measurement leads to better marketing.
That thesis was the thread. We kept pulling. Every measurement problem exposed a data problem beneath it. Before we could model performance, we had to connect touchpoints. Before we could extract meaning, we had to reconcile their definitions. Before we could explain the customer journey, we had to account for what could not be observed.
The more we pulled, the clearer it became: at its core, measurement was—and remains—an infrastructure problem.
The problem kept moving upstream
Marketing data arrives from dozens of platforms, channels, and systems, each describing performance differently. Connecting it was only the beginning. Credible measurement required shared taxonomies, consistent KPIs, and business logic that preserved what every signal meant.
Marketing data is inherently incomplete. Upper-funnel channels like television, radio, and out-of-home have never been observable at the person level. Privacy restrictions, cookie deprecation, walled gardens, and platform proliferation have widened those gaps. We could structure what was observable, but measurement still failed if exposure remained invisible.
That led us to develop our Enriched Data technology. First deployed in 2020, it generates and reconstructs missing customer journeys without cookies or personally identifiable information. It allowed us to preserve person-level measurement even as direct visibility declined.
Over time, those prerequisites became a persistent foundation that could connect marketing data, give it business meaning, fill critical gaps, and preserve the causal relationships needed to understand performance.
The category was accidental. The architecture was purpose-built.
That architecture became the Substrate.
AI revealed what we had built
For most of its history, measurement produced an answer for a person to interpret. AI changes the role of that answer. Models and agents now support planning, optimization, monitoring, and execution as the work happens.
What’s needed is not another interface or model. Its context agents can reason from: structured, complete, and grounded in how marketing affects a specific organization's business outcomes.
Most solutions in the market today begin with a general-purpose model and then look for the data and context to make it useful. We began at the other end. For over two decades, we built the marketing-specific causal understanding required to produce a credible answer. AI arrived after that foundation was already taking shape.
It changed our understanding of what we had built. Measurement was not simply an application. It was infrastructure.
The same causal understanding that explains why performance changed can help people and AI predict outcomes, simulate alternatives, optimize investment, and guide action. AI did not create the value of that context. It revealed how much more the context could enable.
The foundation must outlast the model
Models will improve. Interfaces will change. The intelligence underneath them should not have to be rebuilt each time they do.
The Substrate is open and model-agnostic. Darwin is Marketing Evolution’s native intelligence application, but it is not the platform’s boundary. The same performance context can support ChatGPT, Claude, an enterprise’s own models and agents, or applications that have not yet been created.
Because that intelligence is not tied to one model, it stays with the organization and continues to accumulate as new data arrives. The model is not the moat. The context the model reasons from is.
We followed the problem to its root
We did not pivot from measurement to chase AI. We followed measurement to the fragmented, inconsistent, and incomplete data beneath it—and built the infrastructure required to solve it.
The Substrate already powers Marketing Evolution’s Measurement and Optimization applications and has operated in custom enterprise deployments since 2024. In one deployment, it detected and resolved data-quality issues and anomalies affecting 29% of the dataset before any model touched the data.
It’s now broadly available as the System of Record for Marketing Performance.
We set out to make marketing measurable. We built the foundation that makes marketing intelligent.