Chunking: why passages have to stand alone
A page is retrieved in pieces of a few hundred words. What splitting does to ordinary prose, and the three edits that make a section survive it.
Notes on answer engine optimization, generative visibility, and what actually moves the needle on a page.
A page is retrieved in pieces of a few hundred words. What splitting does to ordinary prose, and the three edits that make a section survive it.
The five faults that keep a good page out of AI answers — each with a by-hand check, a way to run it across a site with or without a terminal, and the fix.
When to phrase a heading as a question, when a keyword label is still the right call, and how to check every heading on a site in one pass.
An element-by-element inventory of what AI crawlers keep, discard, and only conditionally read — plus the parts you can exclude on purpose.
What happens to a page between fetch and answer, the eight signals answer-readiness turns on, and how to check every one of them by hand and at scale.
Reading robots.txt tells you what you intended. Six checks that tell you what actually happens when an AI crawler requests one of your pages.
Every AI bot worth naming in your robots.txt, what each one actually controls, and which ones decide whether an assistant can find you at all.
Which AI bots exist, what each one actually controls, why robots.txt is only half the answer, and how to check whether assistants can reach your pages.
The highest-weighted answer-readiness signal is a passage that answers its own heading in one breath. What it is, how to write one, how to check it.
A walk through what actually gets extracted from a page when an AI system reads it, and which parts of your HTML never make the trip.
Pages that rank fine and still never get cited usually share the same five faults. Each one is fixable in an afternoon.