The ultimate guide to AEO: How to get ChatGPT to recommend your product | Ethan Smith (Graphite)

Key Concepts

• AEO (Answer Engine Optimization) / GEO (Generative Engine Optimization): Optimizing content to appear in the answers provided by LLMs (Large Language Models) like ChatGPT, Gemini, and Claude.
• LLM (Large Language Model) + RAG (Retrieval Augmented Generation): The underlying technology behind answer engines. LLMs summarize search results (RAG) to provide answers.
• Citations: The sources (URLs, videos, etc.) that LLMs use to generate their answers. Getting mentioned in as many relevant citations as possible is crucial for AEO.
• Long Tail: The large number of very specific questions that people ask in chat, which are often not searched for on traditional search engines.
• Share of Voice: The percentage of times your content appears in the answers provided by LLMs for a given set of questions.
• Information Gain: A measure of how much new or unique information your content provides compared to other sources.
• Experiment Design: Setting up controlled experiments to test the effectiveness of different AEO strategies.
• Model Collapse: The degradation of an AI model's performance when it is trained on its own generated content.
• Wisdom of the Crowd: The idea that the average opinion of a large group of people is often more accurate than the opinion of any single individual.

1. Defining AEO and its Significance

• AEO and GEO are essentially the same thing, referring to optimizing for how your content appears in LLM-generated answers. Ethan Smith prefers AEO because "answer" is more narrowly defined than "generative."
• AEO is a significant shift in SEO, comparable to Google's Panda update, which moved SEO from spam to non-spam. This is because it involves a summarization of search results and new inputs.
• Lenny notes that ChatGPT is already driving more traffic to his newsletter than Twitter, highlighting the immediate impact of AEO.
• Unlike traditional SEO, where domain authority is crucial, early-stage companies can win at AEO quickly by getting mentioned in citations.

2. Impact and Examples of AEO

• Ethan disagrees with the notion that "just write awesome stuff" is enough for AEO. He believes anything can be optimized by understanding the underlying systems.
• Webflow is a specific example where AEO efforts have yielded significant results.
• Traditional SEO still works for AEO, but there are additional strategies beyond SEO that are effective.
• Example: For the query "what's the best website builder," simply ranking #1 in Google search isn't enough to win the answer in an LLM. You need to be mentioned as many times as possible across various citations.
• Webflow saw a 6x conversion rate difference between LLM traffic and Google search traffic, indicating the higher quality of leads from answer engines.

3. AEO vs. SEO: Head and Tail Differences

• Head: In traditional SEO, ranking #1 for a keyword like "best website builder" often guarantees a win. In AEO, the LLM summarizes multiple citations, so being mentioned most often is key.
• Tail: The long tail of questions is larger in chat than in search. People ask more follow-up questions and specific queries in chat, creating opportunities to win with niche content.
• Early-stage companies can target these specific, unaddressed questions to gain visibility.

4. Three Levers for AEO

• On-site: Traditional SEO, but with a focus on answering specific questions about use cases, features, integrations, and languages.
• Off-site: Getting mentioned in citations across various platforms:
  • Video (YouTube, Vimeo)
  • UGC (Reddit, Quora)
  • Affiliates (Dotdash Meredith, blogs)

5. The Power of Reddit in AEO

• Reddit is heavily cited in LLMs due to its authentic, community-managed content.
• The effective Reddit strategy is to be an actual user: create an account, state your affiliation, and provide useful answers.
• Spamming Reddit with fake accounts and automated posts is ineffective and leads to bans.
• LLMs are actively policing citations, favoring trusted sources like Reddit.

6. RAG vs. Core Model

• Core Model: The LLM's base knowledge, trained on vast datasets like Common Crawl. Influencing the core model is extremely difficult.
• RAG (Retrieval Augmented Generation): The process of searching for and summarizing information to answer a query. AEO primarily focuses on influencing the RAG process.
• The focus should be on the RAG side because it's more controllable and directly impacts the answers provided by LLMs.

7. Key Considerations for AEO Success

• Recognize that AEO is related to search (LLM + RAG).
• Focus on topics, not just individual keywords. Each page should target hundreds or thousands of related questions.
• Conduct thorough question research to understand what people are asking.
• Optimize citations across various platforms.

8. Actionable Plan for AEO

1. Question Research: Identify questions to rank for by transforming keywords (from search data, paid search, competitor analysis) into questions using LLMs.
2. Tracking: Use an AEO tracker (similar to keyword trackers) to monitor your share of voice and average rank for targeted questions.
3. Citation Analysis: Identify who is currently showing up as citations for your target questions.
4. Landing Page Creation: Create landing pages that answer the main question and all relevant follow-up questions.
5. Off-site Optimization: Implement strategies for each citation group (affiliates, YouTube, Reddit).
6. Experiment Design: Set up controlled experiments with test and control groups to measure the effectiveness of your AEO efforts.
7. Team Building: Assemble a team with SEO expertise and community management skills.

9. Experiment Design: Control Groups and Reproducibility

• Use a control group of questions that you don't touch to compare against your test group.
• Reproducibility is crucial. Repeat experiments multiple times to ensure the results are consistent.
• Most best practices are not correct, so always test and validate your strategies.

10. AEO Tracker and Share of Voice

• AEO trackers are similar to keyword trackers but measure your share of voice (percentage of times you show up) and average rank in LLM responses.
• Consider question variance (different ways of asking the same question) and different surfaces (ChatGPT, Gemini, Perplexity) when tracking.

11. LLM Landscape: ChatGPT, Gemini, Claude, Perplexity

• While the foundational algorithms are similar, the results across different LLMs can vary significantly.
• Focus on the platforms with the most traffic, but don't ignore emerging players.
• There will likely be multiple winners in the LLM space, so optimize for several platforms.

12. B2B vs. Commerce AEO Strategies

• B2B: Citation optimization varies. TechRadar might be prominent. Focus on tracking and attribution beyond last-touch referral traffic.
• Commerce: Shoppable cards and rich snippets are important. Last-touch referral traffic is a good indicator of conversions.
• Early Stage: Focus on citation optimization and long-tail questions. Avoid mid-level SEO efforts.

13. The Importance of Showing Up as a Citation

• Getting cited is crucial because users often open a new tab and search for your brand on Google or directly type in your domain, leading to misattribution of traffic.

14. Should You Allow LLMs to Scrape Your Content?

• It's not your choice whether to play the game. You're playing whether you want to or not.
• Opting out means your competitors will show up instead.
• You can block training on your data while still allowing indexing.

15. Misinformation and Tooling in AEO

• There's significant misinformation about AI and AEO, including claims that Google search is dying.
• Google's slice of the pie stays the same; the pie gets bigger.
• Be wary of extremely expensive tools that perform commodity tasks.

16. AI-Generated Content: The Spam of the Future?

• 100% AI-generated content (no human in the loop) does not work.
• AI-assisted content (edited by humans) is the future.
• There's more AI-generated content on the internet than human-generated content.
• If AI-generated content worked, it would lead to an infinite loop of derivatives and model collapse, diminishing the wisdom of the crowd.

17. Preventing Perverse Incentives in LLMs

• Identify and address perverse incentives like AI-generated content.
• Recognize that LLMs and search are converging.
• Focus on use cases like personalized trip planning and autonomous agents.

18. Help Center Optimization: An Underutilized AEO Opportunity

• Optimize your help center to answer follow-up questions about features, use cases, and integrations.
• Move your help center to a subdirectory instead of a subdomain.
• Improve cross-linking within your help center.
• Create content for the long tail of questions that are not currently addressed.
• Open up your help center to the community to fill in the tail.

19. Conclusion

AEO is a rapidly evolving field with significant potential for businesses of all sizes. By understanding the underlying principles, implementing a data-driven approach, and focusing on providing valuable content, you can increase your visibility in answer engines and drive high-quality leads.