The ultimate guide to AEO: How to get ChatGPT to recommend your product | Ethan Smith (Graphite)
Ethan Smith, CEO of Graphite and SEO expert, discusses Answer Engine Optimization (AEO), the new SEO for LLMs like ChatGPT. He shares tactical advice on how companies can get their products to show up more often in AI answers, emphasizing the significant conversion rate difference from LLM traffic.
Deep Dive Analysis
15 Topic Outline
Defining AEO and GEO: Optimizing for Answer Engines
Impact and Value of AEO Traffic
Early-Stage Startups Winning with AEO
On-Site vs. Off-Site AEO Strategies
Reddit's Role in AEO and Authentic Engagement
LLMs and RAG: How Citations are Used
Key Principles for AEO Success: Topics and Questions
Avoiding Hyper-SEOed Content and Information Gain
Actionable AEO Playbook: Steps and Experimentation
Tracking AEO Performance and Share of Voice
Adapting AEO for B2B, Commerce, and Early-Stage
The Future of AEO, SEO, and Search Channels
AI-Generated Content: What Works and What Doesn't
Dangers of Infinite AI Derivatives and Model Collapse
Help Center Optimization for Long-Tail AEO
5 Key Concepts
AEO (Answer Engine Optimization)
AEO is the process of optimizing content to show up as an answer in Large Language Models (LLMs) like ChatGPT. It focuses on getting content mentioned as many times as possible across various citations, rather than just ranking a single blue link.
RAG (Retrieval Augmented Generation)
RAG is a technique LLMs use to perform a search and then summarize the search results to generate an answer. Most AEO strategies focus on influencing this RAG piece by ensuring content appears in the retrieved citations, rather than trying to influence the core LLM training data.
Information Gain
Information gain is a concept that suggests content is valuable if it says something new that nobody else has said. This principle helps avoid the problem of everyone rewriting each other's content, promoting originality and unique insights.
Share of Voice (AEO)
In AEO, 'share of voice' refers to how often a brand or product shows up in LLM answers across different queries and platforms (ChatGPT, Gemini, Perplexity). It requires tracking not just if you rank, but how frequently you are cited and your average rank across varying responses and question variants.
Model Collapse
Model collapse describes a phenomenon where if AI models are continuously trained on their own derivatives (AI-generated content), they lose the 'wisdom of the crowd' and converge on a single, often incorrect or oversimplified, opinion. This leads to a degradation of quality and diversity in responses.
9 Questions Answered
Traditional SEO aims to get your 'blue link' to rank first in search results, while AEO focuses on getting your content mentioned multiple times as a citation within an LLM's summarized answer, as LLMs often synthesize information from many sources.
Yes, leads from answer engines are significantly more valuable, with one company seeing a 6x conversion rate difference. This is likely due to users being more primed and having higher intent after a conversational interaction with an LLM.
Unlike traditional SEO which takes years to build domain authority, early-stage startups can win quickly in AEO by getting mentioned in citations immediately through Reddit threads, YouTube videos, or blog mentions.
The effective strategy for Reddit is to post useful, high-quality, authentic comments as a real person, stating your name and workplace, rather than attempting to spam with fake accounts, which typically results in bans and deletions.
LLMs primarily use a process called Retrieval Augmented Generation (RAG), where they perform a search to find relevant citations and then summarize those search results to formulate an answer, rather than relying solely on their core training data for real-time information.
Since LLMs don't provide search volume data, one can convert existing search keywords into questions, mine questions from sales calls and customer support interactions, or analyze Reddit discussions to uncover long-tail questions.
Yes, it is generally beneficial to allow LLMs to index your content because if you don't, your competitors will, and you will lose out on potential traffic. You can, however, configure your robots.txt to block LLMs from *training* on your data while still allowing indexation.
No, 100% AI-generated content (without human intervention) does not work well for AEO or SEO, as studies show it performs poorly and can lead to issues like 'model collapse' if LLMs are trained on their own derivatives.
Help center content can be optimized by moving it to a subdirectory (from a subdomain), ensuring good cross-linking, and creating articles for long-tail, obscure use cases and specific questions that customers ask in sales or support, where there might be little existing content.
25 Actionable Insights
1. LLM Traffic Highly Valuable
Prioritize optimizing for LLM traffic as it yields significantly higher conversion rates (6x observed) due to primed user intent from conversational interactions.
2. Leverage Diverse Citations
To rank high in LLM answers, aim to be mentioned across various citation types like Reddit, YouTube videos, and blogs, as LLMs summarize many sources.
3. Optimize for Long-Tail Questions
Map out and answer highly specific, long-tail questions that users ask in conversational AI, as this ’tail’ is larger and offers new opportunities compared to traditional search.
4. Mine Customer Questions
To discover long-tail AEO opportunities, analyze questions from sales calls, customer support, and community platforms like Reddit, as these reflect real user queries.
5. Authentic Reddit Engagement
For Reddit optimization, engage authentically by creating an account, identifying yourself and your company, and providing genuinely useful answers to relevant threads.
6. Create Niche Video Content
Produce YouTube or Vimeo videos for niche, high-value, and potentially ’non-glamorous’ B2B keywords or topics, as there’s less competition and high opportunity for citation.
7. On-Site Product Detail Answers
On your website, create content that thoroughly answers specific follow-up questions about your product’s use cases, features, integrations, and languages to improve AEO.
8. Run AEO Experiments
Design AEO experiments with a control group (questions you don’t optimize) and a test group (questions you apply AEO strategies to) to accurately measure the impact of your efforts.
9. Track AEO Performance
Use AEO tracking tools (an evolution of keyword tracking) to monitor your ‘share of voice,’ average rank, and how often your content appears in LLM answers across different platforms.
10. Prioritize Reproducible Results
Ensure your AEO strategies are reproducible by running experiments multiple times and verifying results, as many online ‘best practices’ are not rigorously tested.
11. Form a Dedicated AEO Team
Assemble a dedicated AEO team, potentially combining your existing SEO team with a community or generalist marketing person skilled in off-site strategies like YouTube and Reddit.
12. Optimize Help Center
Optimize your help center content for AEO by addressing specific product features, use cases, and integrations, as users frequently ask these questions in chat.
13. Help Center Subdirectory
Ensure your help center is hosted on a subdirectory (e.g., yourdomain.com/help) rather than a subdomain (e.g., help.yourdomain.com) for better AEO performance.
14. Improve Help Center Cross-Linking
Improve internal linking within your help center pages to enhance content discoverability and relevance for LLM summarization.
15. Community-Driven Help Center Tail
Actively solicit and incorporate community questions and answers into your help center to cover the long tail of specific, obscure use cases.
16. Early-Stage AEO Focus
Early-stage companies should focus on AEO as they can win quickly by getting cited immediately, unlike traditional SEO which requires domain authority over time.
17. Traditional SEO Still Relevant
Continue applying traditional SEO best practices, such as creating high-quality landing pages for high-volume keywords, as they also contribute to AEO success.
18. Avoid LLM Spamming
Avoid spamming LLMs with low-quality content or automated tactics, as platforms like ChatGPT actively police and penalize such behavior.
19. Optimize for Multiple LLMs
Don’t solely focus on one LLM (e.g., ChatGPT); optimize for several prominent answer engines like Claude, Gemini, and Perplexity, as the landscape is still evolving.
20. Block LLM Training, Allow Indexing
If concerned about LLMs training on your data, configure your robots.txt to allow indexing for answers but block specific training bots, ensuring your content appears without being used for model training.
21. No 100% AI-Generated Content
Do not rely on 100% unassisted AI-generated content for AEO, as it typically performs poorly; instead, use AI as an assistance tool with human oversight and editing.
22. Develop Emotional Intelligence
Develop your emotional intelligence by studying foundational research, as it’s crucial for building relationships and effective communication in both personal and professional life.
23. Learn Persuasion Frameworks
Study persuasion frameworks, such as those by Robert Cialdini, to understand psychological principles that can inform effective growth and marketing strategies.
24. Master Measurement Skills
Cultivate the skill of measuring seemingly unquantifiable aspects in business and growth, as anything can be measured with creative approaches.
25. Embrace Focused Practice
Achieve mastery by engaging in intentional, focused, and intense practice over a significant period, rather than relying solely on innate talent or general effort.
6 Key Quotes
You can get mentioned by a citation tomorrow and start showing up immediately. You can have a Reddit thread. You can have a YouTube video. You can be mentioned on a blog. So early stage companies can win. They can win quickly.
Ethan Smith
Webflow saw a 6x conversion rate difference between LLM traffic and Google search traffic.
Ethan Smith
The whole purpose of Reddit is to post useful high quality authentic comments from real people.
Ethan Smith
Most best practices, most blog posts are not correct. So how do you set up an experiment? You get your questions, you turn tracking on, give it a couple weeks, make your changes, have a test group, have a control group, intervene on the test group, make your changes, see if the chart went up, see if the control group did not, and now you know your particular strategy worked.
Ethan Smith
Google's slice of the pie stays the same. The pie gets bigger.
Ethan Smith
I'm afraid of this world where everything is trained on AI and AI is trained on AI and generating AI and just like nothing is trusted.
Lenny Rachitsky
2 Protocols
7-Step AEO Playbook
Ethan Smith- Identify target questions: Use search data (yours and competitors'), paid search data, and convert keywords into questions using an LLM.
- Track questions: Implement an AEO tracker to monitor your 'share of voice' across different LLMs and question variants.
- Analyze citations: Identify who is currently showing up as citations for your target questions and develop a strategy for each group (e.g., your site, video, UGC, affiliates, blogs).
- Create landing pages: Develop specific landing pages based on the page types that are ranking, ensuring they answer the main question and all relevant follow-up questions.
- Optimize off-site presence: Implement strategies for different citation groups, such as paying affiliates, creating YouTube/Vimeo videos (especially for B2B long-tail), and engaging authentically on Reddit.
- Design experiments: Set up A/B tests with control and test groups for your AEO interventions, tracking changes over time and reproducing successful strategies.
- Build an AEO team: Leverage your existing SEO team/agency for on-site optimization and content, and a marketing/community team for off-site citation optimization (e.g., video creation, Reddit engagement).
Reddit Optimization Strategy
Ethan Smith- Create a Reddit account.
- Identify relevant threads that are part of citations you want to appear in.
- State your real name and where you work (e.g., 'This is [Name] from [Company]').
- Provide a useful and authentic piece of information or answer to the discussion.
- Avoid spamming or creating fake accounts, as the community actively polices this.