What is chunking?
Chunking is the process of splitting a document into smaller pieces before it's indexed, so a search or AI system can retrieve the one relevant piece instead of the whole page.
A full page is rarely the right unit to hand back for a narrow question. Someone asking one specific thing wants the paragraph that answers it, not a 3,000 word article. So the article gets broken up, by heading, by paragraph, or by a fixed length, and each piece is treated as its own separately retrievable object.
This is why one confusing paragraph buried inside an otherwise well written page can get pulled out and misread. Once separated from the page, that paragraph carries none of the surrounding caveats or framing that made it make sense in its original context.
The practical implication for writing: sections that stand on their own, without depending on the paragraph before them to make sense, chunk well. A page that builds an argument gradually, where section four only makes sense after reading sections one through three, chunks badly, no matter how good the argument is once read in full.
Related
- Passage retrievalPassage retrieval is the step where a search or AI system selects a specific relevant passage from an indexed document to answer a query, rather than returning the whole document.
- Retrieval-augmented generation (RAG)Retrieval-augmented generation is the technique of retrieving relevant documents and giving them to a model as context before it answers.
- Answer-first writingAnswer-first writing means stating the direct answer near the top of a page, before the background, story, or caveats that traditionally lead a piece of writing.
Want to know where you actually stand on this? Run a free visibility check or try the free tools.