SEO in the AI Era: What Still Works on a Brand-New Site

The panic take is that AI answers killed SEO. The boring reality: the ranking signals barely moved, but the shape of the win did. You're now competing to be the page an answer engine cites, which rewards clarity and structure even more than before.

Get the mechanics right once#

1. One <h1> per page that says what the page is, not what your company is.
2. A unique <title> and meta description per route — templated, never duplicated.
3. A canonical URL on every page, including paginated and filtered ones.
4. A real sitemap.xml and a robots.txt that doesn't accidentally block your own assets.
5. Structured data (Article, Product, FAQPage) as JSON-LD, matching what's visibly on the page.

1const articleJsonLd = {2  "@context": "https://schema.org",3  "@type": "BlogPosting",4  headline: post.title,5  description: post.excerpt,6  datePublished: post.publishedAt,7  dateModified: post.updatedAt ?? post.publishedAt,8  author: { "@type": "Organization", name: author.name },9  image: [`${appUrl}${post.cover.src}`],10};

Write to be quoted#

An answer engine lifts a passage. Passages that get lifted tend to share a shape: a direct question as a heading, then a two-to-four sentence answer immediately under it, in plain language, with the specifics included rather than promised further down.

• Put the answer before the context, not after it.
• Use concrete numbers and names — “under 30 seconds”, “TLS 1.3”, “89 days” — not “fast” and “secure”.
• Keep one idea per heading so a passage can stand alone when it's excerpted.
• Date your posts and update them honestly; freshness is a real tiebreaker in this category.

The newer files worth having#

Alongside robots.txt and sitemap.xml, an llms.txt at your root gives answer engines a short, curated map of what your site is and where the substantive pages are. It costs an afternoon and removes ambiguity about what you'd like cited.