LLM Optimisation
What Content Performs Best in LLMs (According to Growth Memo)
Kevin Indig analysed 7,000+ AI-driven citations in ChatGPT, Perplexity, Gemini, and more to identify what makes content stand out in generative search results. His findings, published in Growth Memo, challenge traditional SEO wisdom and offer a fresh playbook for content strategy (growth-memo.com).
12 July 2025
12 min read
🔍 1. Depth & Comprehensiveness Over SEO Signals
Classic SEO metrics—backlinks, keyword rankings, domain authority—show minimal correlation with AI citations. Instead, AI tools reward depth, word and sentence count, and readability.
Takeaway: Prioritise long-form, well-structured content that thoroughly explores topics rather than chasing SEO metrics like DR or backlink counts.
🏆 2. What the Top 10% of Cited Content Have in Common
When comparing the most cited content to average pieces, strong patterns emerge:
Lengthy, in-depth text
High Flesch readability
Brand popularity and volume correlate with citations.
✅ 3. Key Priorities for AI Visibility
Create long‑form, comprehensive articles with multiple subsections.
Keep readability high—short sentences, clear language.
Build brand recognition: AI tools favour known entities and strong search volume (reddit.com, linkedin.com).
🚫 4. What Doesn’t Work
Low impact of backlinks, DR, and traditional keyword positions
Mere technical SEO audits won’t move AI citations unless paired with strong content.
✍️ 5. Structuring Content for AI
Use formats that are easily parsed and quoted:
Clear headings and subheadings
Short, tight paragraphs
Bullet lists, tables, Q&A formats
🛠️ LMO7-Tailored Content Strategy Framework
A. Audit & Expand
For existing long-form pages, review and fill gaps—add subtopics, tables, lists, FAQs.
B. New Content Development
Select deep-topic areas and produce comprehensive, structured articles.
Maintain high readability and support with examples, templates, and actionable steps.
C. Brand & Visibility Building
Publish under recognisable branding.
Promote via guest posts or expert quotes to boost visibility in AI crawls.
D. Technical & Discovery
Ensure clean HTML, optimised sitemaps, fast load times to support retrievability in AI searches (akin to LLM crawl-ability).
E. Monitor & Iterate
Track AI citations manually or via tools; perform A/B tests to evaluate impact. Focus on referral traffic from AI citations.
🎯 Example Content That AI Loves
“How-to” guides with step-by-step clarity
Ultimate lists (e.g., “Top 10 tools for X”)
Deep-dive technical explainers with structured data
FAQ pages on niche, detailed queries
🎯 Final Take
To earn citations and visibility in AI-generated answers, think like a publisher, not a traditional SEO. Focus on in-depth, structured, readable content backed by brand awareness. Optimise how LLMs parse your content. Clear headings, rich lists, concise paragraphs and prioritise discoverability via solid site architecture. This isn't ‘SEO Lite' it's content excellence optimised for generative AI ecosystems.
Reference: What content works well in LLMs? by Kevin Indig, Growth Memo, Mar 17 2025
Classic SEO metrics—backlinks, keyword rankings, domain authority—show minimal correlation with AI citations. Instead, AI tools reward depth, word and sentence count, and readability.
Takeaway: Prioritise long-form, well-structured content that thoroughly explores topics rather than chasing SEO metrics like DR or backlink counts.
🏆 2. What the Top 10% of Cited Content Have in Common
When comparing the most cited content to average pieces, strong patterns emerge:
Lengthy, in-depth text
High Flesch readability
Brand popularity and volume correlate with citations.
✅ 3. Key Priorities for AI Visibility
Create long‑form, comprehensive articles with multiple subsections.
Keep readability high—short sentences, clear language.
Build brand recognition: AI tools favour known entities and strong search volume (reddit.com, linkedin.com).
🚫 4. What Doesn’t Work
Low impact of backlinks, DR, and traditional keyword positions
Mere technical SEO audits won’t move AI citations unless paired with strong content.
✍️ 5. Structuring Content for AI
Use formats that are easily parsed and quoted:
Clear headings and subheadings
Short, tight paragraphs
Bullet lists, tables, Q&A formats
🛠️ LMO7-Tailored Content Strategy Framework
A. Audit & Expand
For existing long-form pages, review and fill gaps—add subtopics, tables, lists, FAQs.
B. New Content Development
Select deep-topic areas and produce comprehensive, structured articles.
Maintain high readability and support with examples, templates, and actionable steps.
C. Brand & Visibility Building
Publish under recognisable branding.
Promote via guest posts or expert quotes to boost visibility in AI crawls.
D. Technical & Discovery
Ensure clean HTML, optimised sitemaps, fast load times to support retrievability in AI searches (akin to LLM crawl-ability).
E. Monitor & Iterate
Track AI citations manually or via tools; perform A/B tests to evaluate impact. Focus on referral traffic from AI citations.
🎯 Example Content That AI Loves
“How-to” guides with step-by-step clarity
Ultimate lists (e.g., “Top 10 tools for X”)
Deep-dive technical explainers with structured data
FAQ pages on niche, detailed queries
🎯 Final Take
To earn citations and visibility in AI-generated answers, think like a publisher, not a traditional SEO. Focus on in-depth, structured, readable content backed by brand awareness. Optimise how LLMs parse your content. Clear headings, rich lists, concise paragraphs and prioritise discoverability via solid site architecture. This isn't ‘SEO Lite' it's content excellence optimised for generative AI ecosystems.
Reference: What content works well in LLMs? by Kevin Indig, Growth Memo, Mar 17 2025