About
Who writes this: a pseudonymous technical SEO practitioner who works from logs, APIs, and tests rather than checklists.
I go by g33konaut. That is the whole identity: no headshot, no surname, no bio slide with logos on it. The work either holds up or it does not, and that is a fairer thing to judge me on than a job title.
I got into search sideways. I was the person on the team who could read a server log, so when traffic fell off a cliff and nobody could explain it, the problem landed on my desk. The answer was not in any blog post. It was in the log file, in a redirect chain someone had added months earlier and forgotten. That was roughly the moment I stopped treating SEO as a checklist and started treating it as a system you can inspect.
Ten-plus years later, most of my day still looks like that. I pull crawl data through APIs, join it against logs and Search Console exports, and write the small scripts that let me look at a hundred thousand URLs instead of pretending a sample of ten is representative. I work across the whole surface (technical, content, links, measurement) because the part you ignore is usually where the problem is living.
The method, if it deserves the name: form a hypothesis narrow enough to be wrong, change one thing, wait longer than is comfortable, then check whether the metric that moved is the one you cared about. Most of what I try does nothing. I publish those too.
What I do not believe in
A ranking factor list worth memorizing. Correlation studies on scraped SERPs. Any tool’s score treated as a target. Scale used as a substitute for judgment.
What you get here
Post-mortems with real numbers, methods you can run yourself without buying anything, and a fair amount of “this is what I expected, and here is what happened instead.” If you came for a shortcut, this is the wrong site.
Ground rules
I do not show client data, name clients, or post recognizable screenshots; cases are generalized to the shape of the problem and the order of magnitude. I do not present correlation as cause; when an effect cannot be isolated from an update or from seasonality, the post says so in the post, not in the replies. And I do not quietly delete things I got wrong: corrections are published as their own entry, linked from the original.
Arguments, counter-evidence, and “your query is wrong, here is why” are welcome on X.