LLM SEO

LLM SEO

Guidance for optimising web content to be discovered, retrieved, and cited by large language models and AI-powered search systems (ChatGPT, Perplexity, Google AI Overviews, Claude).

Traditional SEO focuses on ranking in search results pages. LLM SEO focuses on being retrieved and cited in AI-generated answers. Both share foundations (crawlable pages, clear structure, fresh content) but diverge in what drives visibility.

Core Principles

1. Concept Ownership Over Keyword Targeting

Target frontier concepts — topics where the organisation can be first or most definitive. Identify emerging questions in community channels (Reddit, GitHub, Hacker News, Twitter/X) that lack authoritative answers, and create the definitive resource.

Apply the competitor replication test: if a competitor could copy the content tomorrow and match it, the content lacks sufficient depth. Include original data, benchmarks, case studies, or proprietary insights.

2. Depth Over Volume

Substance-rich content outperforms thin coverage. Every page should include:

• Metrics, benchmarks, or original data
• Code blocks, tables, and structured lists
• Worked examples with real numbers
• Clear, consistent terminology (avoid fuzzy synonyms that weaken semantic embeddings)

3. Write for Extraction

AI systems extract self-contained snippets. Structure content so individual sections stand alone:

• Lead with the answer — state the key insight first
• One concept per paragraph — keep sections focused and quotable
• Use structured formats — lists, tables, and code blocks parse cleanly
• Define terms inline — avoid requiring context from other sections

4. Render for Crawlers

Most AI crawlers fetch but do not execute JavaScript. Ensure all content is in the initial HTML:

• Use SSR, SSG, or ISR — never rely on client-side rendering for important content
• Verify by viewing page source (not DevTools Elements)
• Apply Schema.org / JSON-LD markup to reinforce meaning
• Use semantic HTML: <dl>, <table>, <blockquote>, <figure>, <code>

5. Earn Citations Organically

Organic community mentions carry more weight than paid links in training data and RAG indices:

• Share insights in Reddit threads, GitHub discussions, Hacker News, Stack Overflow
• Create open-source resources and tools others can reference
• Build topic clusters with strong internal linking
• Avoid paid links and artificial citation networks

Implementation Checklist

When optimising a page or site for LLM visibility, work through these areas:

Content Quality

• Original data, benchmarks, or insights competitors cannot replicate
• Consistent terminology throughout (no fuzzy synonyms)
• Self-contained sections that stand alone as extractable snippets
• Tables, code blocks, and lists for structured information
• Topic clusters with internal linking between related concepts

Technical Foundations

• Server-side rendered HTML (SSR/SSG/ISR) — no JS-dependent content
• Clean heading hierarchy (H1 → H2 → H3, no skipped levels)
• Schema.org / JSON-LD structured data
• Semantic HTML elements (<dl>, <table>, <figure>, <code>)
• ARIA labels and semantic class names

Crawlability

• robots.txt permits AI crawlers (GPTBot, ClaudeBot, PerplexityBot)
• XML sitemap with accurate <lastmod> dates
• Indexed in both Google Search Console and Bing Webmaster Tools
• Core Web Vitals passing — fast, static HTML/CSS delivery
• No broken links or 404 errors

Authority & Distribution

• Content shared on high-signal channels (Reddit, GitHub, HN, Twitter/X)
• Open-source resources or tools that earn organic references
• Active engagement in relevant community discussions
• Regular content refresh cycle (30/90/180-day cadence)

Measurement

• Citation checks in Perplexity, ChatGPT, Google AI Overviews, Claude
• Analytics tracking referrers from AI platforms
• Index coverage monitoring in Search Console / Bing Webmaster
• Community mention tracking (Ahrefs, Mention, Semrush)

Content Refresh Cadence

Stale content loses retrieval potential as models re-crawl:

Interval	Action
30 days	Review performance, fix broken links
90 days	Update underperforming pages, expand successful content
180 days	Archive outdated pages with redirects, audit full catalogue

Always update lastmod timestamps when refreshing content.

Additional Resources

Reference Files

For detailed guidance on specific areas, consult:

• references/technical-implementation.md — HTML structure, rendering strategies, crawlability configuration, and extractable snippet patterns
• references/citation-authority.md — Concept ownership strategy, finding frontier topics, organic citation building, and high-signal channel guidance
• references/strategy-comparison.md — Traditional SEO vs LLM SEO comparison table, mindset shifts, content refresh cadence, and measurement approaches

Source

Distilled from Vercel's guide on adapting SEO for LLMs and AI search.