Our Evidence Standard
Every recommendation on AEOing carries one of three labels. Proven comes from a primary source such as a platform’s own documentation. Practice is sound, common practice that platforms have not confirmed. Experimental is unproven.
The three labels
- Proven
- A search or AI company states it in its own documentation or announcement, and we link to it.
- Practice
- Consistent with how the systems work and widely used, but not confirmed by the platforms. We explain the reasoning.
- Experimental
- A proposal or hypothesis with no confirmed effect. We say so plainly and we do not promise results.
What we checked (22 October 2026)
Google’s AI features documentation says a page must be indexed and eligible to be shown with a snippet, with no additional technical requirements, and that you do not need new machine-readable files, AI text files or special markup.
We also read the live robots.txt of 15 sites across SEO publishers and tool vendors and AI and developer companies. Of the 14 whose robots.txt we could read, none blocked AI crawlers site-wide. Eight of the 15 published an llms.txt file. Google and OpenAI did not. This describes behavior, not proven benefit.
This site’s own setup
- robots.txt allows both AI search and training crawlers. View it.
- We publish llms.txt as an experiment. It is labelled unproven.
- Structured data (Organization, WebSite, Article, BreadcrumbList, DefinedTermSet) always mirrors visible content.
- Every page shows a last updated date. We revise labels when platforms change their guidance.
Corrections
If something here is wrong or out of date, tell us through the contact page. We will correct it and update the date.