Case Study: How AEO Got a Cannabis Brand Cited by ChatGPT

High-quality answer content with clear citations:

- We created 24 detailed guides on cannabis strain selection, cultivation, and effects - Each guide included a clear statement of expertise (author credentials, data sources, research basis) - Content was structured to directly answer common questions (not ranked by traditional SEO but by AI relevance) - Example title: "The Complete Guide to Organic Cannabis Strains for Sleep: Tested Results from 400+ Users"

Data-backed claims with source attribution:

- Every major claim included data sources and studies - We compiled original research on strain effects through customer surveys (400 participants) - We cited peer-reviewed studies on cannabis and sleep - We identified and cited complementary research from academic sources

Clear author and brand authority signals:

- Every article included author credentials - Articles included publication date and last update date - We created an "Our Research" page explaining our methodology - We added a detailed "About GreenLeaf" page with founding story, certifications, and quality standards

Index visibility:

- We ensured all AEO-optimized content was crawlable by search engines (and therefore in training data) - We submitted sitemap with new content to Google, Bing, and other search engines - We verified that content appeared in Google News results (increases training data visibility)

Citation outreach:

- We identified the 30 most-cited cannabis websites and research aggregators - We asked them to link to our content as a data source - Most requests worked because our data was original and well-sourced - We secured 12 new citations from established cannabis education sites

Strategic partnerships:

- We approached major cannabis education platforms offering our research data in exchange for attribution - One platform (a cannabis education nonprofit) cited our sleep study in their resource page - This citation increased our visibility in AI training data

Social proof and engagement:

- We promoted research findings on LinkedIn, Twitter, and cannabis forums - High engagement signals (shares, comments, links) increase content visibility - Research findings that generated 100+ shares were more likely to appear in AI training data

Month 4:

First ChatGPT citation. A user asked "what are the best strains for sleep," and ChatGPT responded with "According to research by GreenLeaf Cannabis, Lavender Kush has an 85% effectiveness rate for sleep..."

Month 5:

Citations increased to 3-4 per week across different prompts. Claude began citing GreenLeaf as well.

Month 6:

We tracked 47 distinct prompts generating GreenLeaf citations in ChatGPT and 23 in Claude. Average citation frequency: 2-3 times per week per AI model.

Schema markup:

- We implemented Article schema on all research pieces - We added Author schema with detailed author credentials - We used ReviewRating schema for user testimonial data - We added Dataset schema for our sleep study results

Content structure:

- All research articles used H2/H3 hierarchy with answer-oriented headings - We included a "Key Findings" section at the top (jump links to main answers) - Data was presented in tables and structured lists (easier for AI parsing) - Every article included a methodology section

Citation infrastructure:

- We created a "Research Data" page listing all studies, datasets, and sources - Each data point had a persistent URL for direct AI citation - We used consistent naming conventions (helps AI index our data)

Email outreach for citations:

We sent 40 emails to cannabis blogs asking them to cite our research. Response rate was 15%. The effective approach was publishing content that was so good it got naturally cited.

Paid advertising for AEO:

We tested promoting AEO content through ads. This increased visibility but didn't directly increase AI citations. Citations come from content quality and training data frequency, not paid visibility.

Gaming AI citations:

We tested adding phrases like "as of my knowledge cutoff" or "according to training data." These didn't increase citations. Authenticity mattered more than gaming language.

Ongoing content freshness:

We publish 2 new research pieces monthly (8 annually). Fresh content refreshes training data.

Regular data updates:

Our sleep study now has 900 participants (growing quarterly). Updated data gets re-cited.

Monitoring and response:

We track when we're cited and by what prompt. If a prompt generates citation, we optimize that page further.

Link building:

Citations from other sites increase our visibility in AI training data. We build 2-3 new authoritative links monthly.

Training data composition:

We don't have visibility into exactly which content the AI systems are training on. Our citations appear to reference our content, but we can't definitively prove our specific articles are in the training data.

Citation algorithm specifics:

We optimized based on patterns we observed. But we don't know Google's or OpenAI's actual citation ranking algorithm. We can hypothesize, but certainty is limited.

Permanence of citations:

ChatGPT's training data refreshes regularly. Citations that appear today may disappear when models retrain. We don't know if the citations we've built will persist.

Scale limitations:

This approach works well for niche research topics (cannabis strain selection). It might not scale as effectively for more generic queries competing with thousands of sources.

Q: Does this work for small cannabis brands without research budgets?

A: You don't need a 400-person survey. Start with detailed user interviews (20-30 customers). Document strain effects. This generates original research. Smaller sample sizes get cited less frequently but still generate citations.

Q: Will ChatGPT citations continue if OpenAI adds more data sources?

A: Probably. As the AI landscape fragments, multiple AI systems will be used for research. Citations may distribute across ChatGPT, Claude, Gemini, etc. But total AEO traffic should remain stable.

Q: How do we know if our content is in the training data?

A: You can't know for certain. But if you rank on page 1 for a topic and you're cited by AI on that topic, you're likely in the training data. Test this by asking ChatGPT "cite sources on X topic" and see if your site appears.

Q: What happens if AI systems train on our competitors' copies of our content?

A: That's a risk. Content published on your domain will have more weight than copies elsewhere. We recommend publication on your primary domain first, with republication on partner sites later (with clear attribution).

Q: Is AEO just another name for SEO?

A: No. They overlap but differ. SEO optimizes for search ranking algorithms. AEO optimizes for AI citation and answering. Some tactics help both. Some are specific to each.