AI pulls from the page with the clean answer because it extracts what it can confidently lift, and a clearly stated, self-contained, specific answer is liftable while a vague one isn’t. When an AI system answers a question or cites a source, it’s looking for a passage it can quote or paraphrase without risk of getting it wrong. A page that states the answer plainly hands it exactly that, so the system reaches for it. A page that gestures at the answer through hedging and scattered fragments offers nothing it can grab with confidence, so it moves on, even if that page is technically authoritative.
Clarity and self-containment do the work. An answer written as a complete statement, “X is Y because Z,” can be lifted whole and still make sense, which is what extraction needs. The page that wins isn’t necessarily the better-researched or higher-ranked one; it’s the one that says the thing directly in a form that survives being quoted out of context. Specificity helps too: a concrete number, name, or definite verdict is extractable in a way that “it varies” and “many factors are involved” simply are not.
The vaguer page usually fails for one of three reasons. It hedges, wrapping the answer in so many qualifiers that no single clean statement exists to pull. It scatters, spreading pieces of the answer across several paragraphs so the system can’t assemble them reliably. Or it stays abstract, talking around the topic without ever committing to the specific answer the question asked for. In each case the information might be present, but it isn’t in a shape an AI can confidently extract, so a clearer competitor gets cited instead.
The mistaken assumption is that AI just picks the highest-ranked page. Extractability often matters more than position, which is why a smaller page with a sharp answer can get cited over a stronger one that buries it. To be the page that gets pulled, state your answer clearly and self-contained, near the top, in a sentence that holds its meaning even when quoted alone, then add the depth beneath it.