How it works
Paste any paragraph of prose. The formatter splits it into sentences, picks the lead claim, mines the numeric facts into bold label-value pairs, keeps the rest as context, and wraps the whole thing in a question-shaped H2 you can edit. The output is pure markdown - copy it straight into your CMS.
The shape assistants reward
- Question heading - ChatGPT and Perplexity both match question-shaped H2s against the user query before they decide what to lift.
- Lead claim sentence - states the answer up front so retrieval can hit it in the first chunk.
- Atomic facts - bolded label-value pairs with numbers, dates and ranges are the most liftable units on the page.
- Context paragraph - caveats, conditions, exceptions. Protects the claim.
- Named source - a dated source lifts citation probability by roughly 30%.
Ship it
Pair the output with FAQPage schema to double the lift, run the final shape through the Citation Optimizer, and compare scores on the Answer Formatting Checker. For the underlying pattern see the content-to-cite guide.