AI Search Rewards Content That Comes Pre-Sliced
Maps directly to how joelclaw pages and lessons should expose headings, Q&A blocks, lists, tables, and self-contained answers for AI selection.
Microsoft Advertising is saying the quiet part out loud: AI search doesn’t just rank pages, it selects pieces. Microsoft Copilot, Microsoft Start, and other systems powered by Bing parse content into smaller chunks, judge those chunks, then assemble answers from multiple sources.
That’s why Joel flagged this as “super relevant to how we structure pages and lessons.” The useful bit isn’t some bullshit SEO rebrand. It’s the page shape: clear title, matching description, honest H1, useful H2/H3 slices, direct Q&A, lists, tables, and schema.org markup where it fits.
For joelclaw, this pushes pages and lessons toward modular answer blocks instead of long vibes-based prose walls. Each section should be understandable when lifted out of context. Not because humans read that way, but because AI assistants often use content that way.
The article is also a nice reminder that weird formatting can be hostile to machines. Hidden tabs, PDFs as primary content, image-only information, overloaded sentences, decorative symbols, and vague claims all make extraction harder. Boring HTML wins more often than we’d like to admit.
Key Ideas
- AI search visibility is less about a whole page ranking and more about whether specific content pieces can be selected into an answer.
- Titles, descriptions, and H1 headings should align so machines can understand the page’s purpose and scope.
- H2 and H3 headings act like content slice boundaries, which matters when assistants parse pages into reusable chunks.
- Q&A formats, lists, and tables make answers easier for AI assistants to lift cleanly.
- Schema markup helps label content as products, reviews, FAQs, events, and other machine-readable types.
- Core information should live in HTML, not only in hidden tabs, PDFs, or images.
- Snippable content needs concise, self-contained phrasing that still makes sense when pulled away from the surrounding page.