Chunk optimization
Chunk optimization is shaping a page's headings, paragraphs and lists so the pieces an AI system pulls out are the ones you'd choose.
Chunk optimization is the work of shaping your page so that when an AI system cuts it into pieces, each piece is one you'd be happy to see quoted. You arrange headings, paragraphs and lists so the cuts fall in sensible places and every section carries a full answer.
It's one of the most direct changes you can make on your own site, because it's all in your hands: once a platform reads a page, the page's shape decides how usable it is.
How to do it
Start with the questions the page should answer. Then:
- Give each question its own heading, worded the way people ask it
- Answer it in the first sentence or two under the heading
- Keep each section to one subject, and split long ones
- Name the subject in each section, instead of saying "as above"
- Put prices, sizes and other key facts in the page's text, where a bot that skips images and scripts still finds them
- Use lists where the content really is a list, like steps or options
An example
Fernhill Coffee's FAQ page had 900 words of mixed answers under a single heading. Fernhill rewrote it with a heading for each question: how long beans stay fresh, how to pause a subscription, where it delivers. Each answer now opens with a direct sentence.
Nothing new was added. The page simply went from one block that systems had to cut at random to eight sections that each answer one question.
Don't overdo it
Write for people first. A page broken into dozens of tiny sections with keyword headings reads badly, and readers leave. The same changes that help a system pull out a clear answer usually help a person skim to theirs.
In AnswerOrbit
Run a page check on a page and a prompt. It tests the page's headings, whether the prompt appears in one, and whether the page opens with an answer, next to the pages the platforms cite for that prompt.