Experiments & research
A working log of the AI and SEO ideas I'm exploring, prototyping, and testing — shared openly as I learn. Nothing here is a finished product; it's the messy middle of building.
SERP-informed metadata generation
TestingReverse-engineering the top-ranking titles for a keyword, then generating meta titles and descriptions engineered to beat them on click-through. Studying which power words and structures actually win the SERP.
Content brief automation from a single keyword
BuildingTurning one keyword into a full SERP-informed content brief and first draft in under 30 seconds — so the human editor only reviews, never starts from a blank page.
Executive growth reports from raw GA4 + GSC data
BuildingConnecting analytics sources and having an AI layer normalise the numbers into a plain-English weekly report — compounding wins, decaying pages, and a ranked next-action list.
Document-to-metadata pipeline
ExploringParsing PDFs and DOCX files to derive page intent and generate metadata in bulk — useful for migrating large content libraries without manually reading every page.
Early-warning radar for rising search trends
ResearchingMonitoring a category for emerging queries and staging them by Early / Rising / Peaking, so content gets published before the moment peaks rather than after.
Follow the experiments
I share what shipped, what failed, and what I learned in the newsletter — plus early access to new prototypes.