I've been improving something recently that I haven't talked about in quite a while.
Some of you might remember my Progressive Optimization system from Optimizers Club.
I developed it years ago as a framework for approaching SEO as an ongoing process of measurement, experimentation, and decision-making, rather than simply working through a checklist of things you're supposed to optimise.
The basic idea was to divide a website into manageable groups of similar pages, whether by category, search intent, or page template.
I'd collect performance data alongside information about the pages themselves, then select specific optimizations to test against groups of pages left unchanged as controls.
Rather than attempting to optimize an entire website in one enormous project, improvements could be implemented progressively, with the results of each experiment helping inform what to do next.
Anyway, I recently decided to give the framework to AI and ask it to critically evaluate the approach against the realities of search marketing in 2026.
Its assessment surprised me a little.
It suggested that the framework might actually be better suited to modern search marketing than it was when I originally designed it.