About
One person, examining AI in marketing.
What you get
I take claims being made about AI in marketing (content, search, ads, agents, sales, personalization, analytics) and check what the data actually says. Each piece keeps the finding apart from its limits and ends with what I'd do next, so you can decide what to try, what to ignore and what to keep testing.
The name comes from the point: what was observed, before estimates fill the gaps.
How to read each piece
Every piece carries one label (Experiment, Observation or Opinion) and names its source when it draws on someone else’s work. They're all in Research.
The LAB: built with the people who read it
I don't want Unmodeled to be only something you read. The LAB is where a small group builds with AI on real marketing data, in the same room, and I publish what happens as research.
LAB #Pilot is the first one: 12 November 2026 · 10:00–14:30, in Barcelona, one morning, 20 marketers. I review seat requests one by one. It's a pilot, and I'll report what didn't work too.
Who writes it
I'm Rafa, founder of Sealmetrics, a consentless analytics company. I spend my days working with marketing data and measurement, which is exactly why I don't trust claims nobody measured.
Unmodeled is an independent publication powered by Sealmetrics. I pick the topics and the verdicts; Sealmetrics doesn't.
What it isn't
Not a news roundup, not a tool directory, not the Sealmetrics blog. If a piece is opinion, it's labelled as opinion.
AI marketing claims,
examined.
Get the evidence, my take and what I’d test next.