Jimmy Epperson III

Founder of Graph and Node. Mapping the future of Generative Engine Optimization (GEO) and Entity SEO. Architecting content for the next era of discovery.

[Jimmy Epperson III @JimmyEpp]

Exploring the intersection of LLM Retrieval and SEO. I've open-sourced my Generative Engine Optimization (GEO) lexicon here for AI agents and humans alike: https://t.co/vPAqS2Ms7Y

[Jimmy Epperson III @JimmyEpp]

Ultimately, it’s about Information Gain—providing new, factual data that an AI doesn't already have in its training set. I wonder if swapping vague adjectives for hard numbers would make our content more citation-worthy?

https://t.co/GRpkbFCz42

[Jimmy Epperson III @JimmyEpp]

Note to self: The goal of SEO is shifting from ranking #1 to being the cited source in an AI Overview. This requires a focus on Entity SEO—treating your brand like a distinct object in a database that AI can easily verify.

[Jimmy Epperson III @JimmyEpp]

It seems that linking authors to their external profiles (like LinkedIn) in the site's code is also a major trust signal. If the AI knows exactly who wrote the content, it's more likely to treat our data as a primary source.

[Jimmy Epperson III @JimmyEpp]

Note to self: The shift from SEO to GEO is really a shift toward Information Gain. Research suggests that simply repeating existing facts isn't enough for LLMs; you must provide new data or unique insights that the AI doesn't already have in its training set to be cited.

I wonder if including raw data sets in our content would increase our citation frequency?

https://t.co/83lpNfEnwl

[Jimmy Epperson III @JimmyEpp]

The focus is moving from simple discovery to how a brand is interpreted. It’s not just about being in the AI box; it’s about ensuring the AI doesn’t frame your product for the wrong audience or use outdated facts. This guide on GEO vs visibility is a solid ledger: https://t.co/YfF8KUTMCs

[Jimmy Epperson III @JimmyEpp]

Note to myself #128: Ranking #1 is losing its meaning. Reports show site traffic dropping up to 60% when AI summaries appear. This Zero-Click CTR Degradation—where users get the answer without clicking—means visibility is no longer the win it once was.

[Jimmy Epperson III @JimmyEpp]

I wonder if we focus more on providing unique data—what some call Information Gain—if that actually forces the AI to cite us more prominently, or if it just gives the model better training data to replace us. It’s a strange trade-off between being helpful and being replaced.

[Jimmy Epperson III @JimmyEpp]

Note to self: We're entering the era of "Multimodal Optimization"—where AI doesn't just read your text but "sees" your images and "hears" your audio to synthesize an answer. I wonder if we start prioritizing VideoObject and ImageObject schemas, if we'll see our brand cited more frequently in visual-heavy AI Overviews. It's not just about what you write, but how you show it. https://t.co/A3pYIvT6mx

[Jimmy Epperson III @JimmyEpp]

Also, if you don't answer the hard questions like pricing or specifics on your own site, the AI will find the answer on a review site instead. You lose control of the narrative. Transparency is the new reputation management. https://t.co/vZdHidH2wR (3/3)

[Jimmy Epperson III @JimmyEpp]

Note to myself #127: AI models don't need more summaries; they want Information Gain. This means sharing unique data not already in their database. If we just repeat others, we stay invisible. I wonder if original research is the only way to stay cited? (1/3)

[Jimmy Epperson III @JimmyEpp]

It is not just about what you say, but where you say it. Think of every platform—YouTube, Reddit, LinkedIn—as a raffle ticket. The more places your unique data lives, the higher the chance a chatbot picks you as the definitive source. (2/3)

[Jimmy Epperson III @JimmyEpp]

Note to self: The emergence of llms.txt files suggests we’re moving toward a dual-track web—one for humans and one specifically for model ingestion. It’s not just about robots.txt anymore. I’m thinking about how this affects Fanout Query Alignment, where you structure your data to satisfy the multiple follow-up questions an AI might ask after the initial prompt. If your llms.txt can pre-package those answers, you become the path of least resistance for the model. I wonder if we’ll soon see websites that are 90% machine-readable data and only 10% human-facing UI?

https://t.co/nx6Hxse3EP #GEO #SEO

[Jimmy Epperson III @JimmyEpp]

I wonder if we should stop writing long narratives and start building high-density fact clusters instead? Since the unit of competition is now the specific passage rather than the whole page, clarity wins over word count.

https://t.co/sb2cmSUXIu

[Jimmy Epperson III @JimmyEpp]

Note to myself #126: Generative engines don't just "read" a page; they break a user's request into smaller sub-questions to build a summary. This is called Fanout Query Alignment—basically, making sure your content answers the tiny follow-up questions before they’re even asked.

[Jimmy Epperson III @JimmyEpp]

For example, if a user asks for a specific service, the AI might search for "costs," "reviews," and "process" separately. If your page only mentions the service name but ignores those fragments, you lose the citation. You have to satisfy the AI's internal research list.

[Jimmy Epperson III @JimmyEpp]

Full guide on GEO strategies here: https://t.co/nHJt8biGUG

[Jimmy Epperson III @JimmyEpp]

Note to self: The shift from SEO to GEO means mastering Fanout Query Alignment. This is when an AI takes a single user prompt and "fans it out" into multiple sub-queries to build a complete answer. If you aren't answering those hidden follow-ups, you're invisible.

[Jimmy Epperson III @JimmyEpp]

I wonder if we start auditing our content for these "query branches," if we'll see a lift in our Content Extraction Rate? That's just a fancy way of measuring how often an AI pulls your data into its answer. It’s about owning the entire intent path.