The document
Being the source an AI answer cites, when nobody clicks through
AI summaries are answering questions on your subject using your pages, and most of that traffic never arrives. This is about being the source that gets used, and about measuring something when the click has gone.
Six claims are tested against documentation rather than repeated. Five of the things most commonly recommended turn out not to be real, and the document says which and why.
What is inside
Three things to get right, and five myths to stop chasing
The amount of confident advice about optimising for AI search is large and the amount of it supported by anything published is small. Every claim here carries its source, and the claims that failed are listed rather than quietly dropped.
Exposure
Whether your pages are eligible to be used at all. Indexing, snippet settings and the controls that genuinely affect whether a model can read you.
Extractability
Whether an answer can be lifted cleanly from your page. Structure, the first paragraph, and why a page that buries its answer gets passed over.
Dependence
How much of your traffic relies on queries an AI answer can finish. The pages most at risk, and what to do about them before the decline shows.
The five myths
Including the special file, the AI-specific schema and the hidden ranking factors. Each one tested, each one sourced, each one dismissed with the reason.
Answer in the first paragraph
The single structural change with the most support behind it, written out with a before and an after.
Measuring without clicks
Impressions holding while clicks fall, click-through by query rather than sitewide, branded versus non-branded, and enquiries per page.
How it works
Check exposure, fix extractability, then watch dependence
The order matters. There is no point restructuring a page that is not eligible to be used, and no point worrying about dependence on pages nobody was reaching anyway.
By the numbers
What you are getting
Before you ask
Questions people send us about this one
Can I stop AI summaries using my content?
Partly, and the document is specific about which controls exist and what each one costs you. Most of them remove you from ordinary search results as well, which is usually a worse trade than the one being avoided.
Is there a file or tag I should be adding?
No. That is one of the five claims tested, and nothing published supports it. The document explains where the idea came from and why it keeps being repeated.
How do I know if this is already costing me traffic?
Impressions holding steady while clicks fall is the signature, and there is a section on confirming it properly rather than reading a seasonal dip as a trend.
Does this replace normal SEO?
No. Everything that makes a page eligible to be cited is something that already made it rank, which is the reassuring part. The new work is structural rather than strategic.
