The lede paragraph is doing two jobs
Your opening paragraph is the candidate text both Google's featured-snippet extractor and LLM retrievers use to summarise the page. It needs to fully answer the question without requiring the reader to scroll. Aim for 40–80 words, declarative tone, no anchor sales language.
If you can't answer the question completely in the first paragraph, the page either covers too broad a topic (split it) or hasn't been edited tightly enough. Across tested engagements, pages rewritten to lead with a complete answer paragraph saw both featured-snippet wins and AI-citation lifts within 4–8 weeks.
Depth, not length
The 'minimum 1,500 word' SEO advice is dated. What matters is whether the page covers the topic with enough depth to satisfy intent and surface differentiated information. A genuine 800-word answer outperforms a padded 2,500-word version on both ranking and citation share.
Add specifics: numbers, dates, named tools, named methodologies, original observations from your work. LLMs preferentially cite content with concrete, attributable details over content full of hedged generalities. Search engines do too, increasingly.
Author and source signals
Show a named author with a real bio, credentials, and links to other work or social profiles. Include a 'last updated' date and update it when material changes. Cite your sources — both inline links and a 'sources' or 'further reading' block at the bottom.
These E-E-A-T signals were always SEO best practice. In 2026 they're disproportionately important for AI citation: LLM retrievers use author signals as a trust filter, and pages with anonymous bylines get downweighted in citation selection across every major assistant.
Schema and structure
Use Article schema with author, datePublished, dateModified, and publisher. Add FAQPage schema where there are genuine questions. Add HowTo schema where the page is genuinely procedural. Don't fabricate schema that isn't reflected in the visible page — both Google and LLM retrievers downweight content with schema that doesn't match the rendered DOM.
Use clear H2 subheadings that signal what each section answers. Subheadings function as boundaries for both Google passage indexing and LLM passage extraction. Vague subheadings ('Learn more', 'Why us') get ignored; specific subheadings that match query intent ('How much does X cost', 'When to use X vs Y') get extracted.