A content chunk is a self-contained passage, typically a single section of roughly 100 to 300 words, that answers one question without depending on the rest of the page. It is the unit AI engines work in: retrieval and citation happen at the passage level, not the page level, so an engine lifts the chunk that best answers a sub-query wherever it sits, and may cite several chunks from different pages in one answer.
Chunking comes from how retrieval-augmented systems index the web. A page is split into passages, each embedded and stored, and when a query fans out into sub-questions, the engine matches each sub-question to the closest chunk. A passage buried in a wall of prose, or split awkwardly across sections, is harder to retrieve cleanly, which is why answer-first, self-contained sections outperform undifferentiated text.
Read the chunk, not the page, as your unit of optimisation. A short page that fully answers one question can be a perfect chunk; a long page that answers eight questions is really eight chunks competing on their own merits. The practical work is to give every real sub-question its own liftable section, and the practical way to know it worked is your measured retrievability and citation coverage per engine, since no engine reports chunk-level selection directly.