AI Visibility
Passage extractability
How easily an answer engine can lift a self-contained block of text that directly answers a question, with the question and answer co-located.
By Shimon Carroll, Founder, SEO for AI Agents · Last updated
Passage extractability is the property of content being structured so a model can grab a single block that answers a specific question without needing the rest of the page. Answer engines reason at the passage level: they retrieve and rank chunks, then compose an answer from the best ones. If your answer is spread across three paragraphs under a vague heading, it is hard to extract; if it is a tight block right under a question-shaped heading, it is easy.
The practical pattern is a question as an H2 or H3, immediately followed by a direct answer of roughly 40 to 170 words, then optional supporting detail. Lead with the answer, not the build-up. Keep one idea per passage. Use plain declarative sentences, because the engine is going to quote or paraphrase them. Tables and short lists are highly extractable for comparison and step questions.
This is the same instinct that won featured snippets, applied to a world where the snippet is composed rather than copied. The reason it matters more now is that an extractable passage is the difference between being the source the model cites and being a page the model skims and discards. We score the structure of a site's most important pages against this pattern.