Prescient AI Research

Research

Original research on the math behind marketing measurement. We publish the methods, the failure modes, and the numbers.

Papers 01
Areas Methodology
What we publish

We treat marketing measurement as a science problem, not a dashboard problem. Most MMM failures aren't statistical — they're structural. You can't tune your way out of the wrong model.

01

Mechanism, not correlation

We model marketing as a dynamical system with latent state, not a linear decomposition of observed outcomes.

02

Identifiability first

A reliable answer requires a model that can recover the right answer in principle. We start with what is identifiable, then build out.

03

We publish what doesn't work

Negative results, structural limits, and benchmarks against open-source baselines — written up the same way we write up wins.

Archive

1 published

Coming soon

More papers are coming. Subscribe and we'll send each one the day it publishes.

New papers,
in your inbox.

When we publish, you get the paper, the methods, and the data. No newsletter chatter. No roundups. Just the research.

One email per paper, never more. Unsubscribe in one click.

Want to see the methods in production? Book a demo