The proving ground

A method you have not run yourself is somebody else’s opinion.

The framework at the root of this method was not mine. It was already winning inside American agencies when a partner handed it to me with one instruction: copy it.

Copying is where I started, not where I stopped

If a framework works, the interesting questions are why it works, where it breaks, and what it becomes when you push it past the point its authors stopped at.

So the year went on testing it rather than trusting it.

The year, as a record

What was tested, and what happened to it.

ReadThe most respected playbook in print, cover to coverKept as the base
RanThe two most heavily marketed AI platforms, on live client workDiscarded
RanDozens of other tools, each on a real engagementMost discarded
RebuiltEvery finding that held, back into the frameworkKept or replaced
BuiltA 32-page agency site from an empty folderThe stress test

The research log · no client, no Google numbers, none invented

The stress test

Theory that has never carried weight is decoration. So the refined method got the hardest test available: a complete 32-page agency site built from an empty folder.

Full silo architecture, every page with a defined job, every internal link documented with its reason. No template, no inherited content, nothing to hide behind. The build surfaced what only building surfaces, and all of it went back into the map.

Folded paper, the light raking across the creases
Built from an empty folder, to find out what survived
Dark crumpled paper, the creases catching a low light

Nothing here runs on theory.

The method was tested and refined before anyone was charged for it, and it has carried client work across three countries since.

You can buy this piece on its own. It is the whole method service. See the case studies