Blog
Short pieces written between editions of the AI Shortlist Index: what the engines did, what we measured, what a vendor should change. No announcements.
Mention rate, position, consistency and the source map, and why one run per prompt is not a result. Field notes from the first AI Shortlist Index edition.
By Ryan York · 2026-10-03
How the AI Shortlist Index designed its 25-prompt set per category, the rules behind it, and how a property-ops vendor builds one for its own category.
By Ryan York · 2026-09-26
What makes a comparison table usable by an engine reading HTML, what the first AI Shortlist Index run cited instead, and one Property Meld row rewritten.
By Ryan York · 2026-09-26
What webpossible.com's Markdown twins are, how the build generates them with a canonical to the HTML page, and why we have no evidence engines read them.
By Ryan York · 2026-09-26
Why a screenshot of one ChatGPT answer proves nothing about AI visibility, what a defensible claim needs, and how to read a vendor's number-one post.
By Ryan York · 2026-09-26
Where run-to-run variance in AI answers comes from, how the AI Shortlist Index bounds it with three runs on two days, and how to tell movement from noise.
By Ryan York · 2026-09-26
Why the AI Shortlist Index publishes its prompt text, protocol and policy, what that costs in gaming and copying, and how anyone can reproduce a row.
By Ryan York · 2026-09-26
What ChatGPT, Perplexity and Gemini pull from a property-ops vendor page, what breaks extraction, and a before-and-after rewrite of one paragraph.
By Ryan York · 2026-09-03
Published by WebPossible. The long-form versions live in the methodology pillar; the numbers live in the measurement framework.