Most writing about AEO and GEO stays theoretical. Here is what I actually did, what happened, and what I think it means.
The First Result
In early 2024, I was managing digital marketing at JLLT (JLL Technologies), a proptech startup under JLL branding. Target articles for several enterprise software category terms were either unranked or buried beyond page two. I developed a two-prompt AI content workflow — using a first prompt to generate a comprehensive answer to a specific “what is” query, and a second to restructure and compress it for answer-first clarity and entity precision.
One article, targeting a facilities management software category query, moved from unranked to position three on Google. In April 2024, it earned a citation in Google’s AI Overviews. Google officially launched AI Overviews in May 2024. The citation predated the launch.
I did not predict that outcome specifically. What I did was write for the way LLMs retrieve and cite information: a clear definition, structured subpoints, entity specificity, and no hedging or filler. It turned out that is also what Google’s AI Overview system selects.
The Second Result
I now provide SEO and digital marketing consulting for Bigfoot Outfitters, an outdoor recreation and lodging company operating on the Ocoee River corridor in Western North Carolina. The business is niche, the query clusters are local, and the brand had almost no online presence when I started.
I built content targeting four distinct topic clusters: guided fishing trips and rainbow trout on the Ocoee, white-water rafting and lower Ocoee logistics, cabin rentals for rafting visitors, and general area information around Benton, TN. The content followed the same structural logic as the JLLT work — answer-first, entity-specific, geographically precise, no filler.
Content across multiple clusters is now cited in Google AI Overviews, ChatGPT, and Perplexity. Not just one article. Not just one platform. A pattern across topic clusters and across three different AI systems.
What Actually Made the Difference
I am careful about turning two results into a grand theory, but there are consistent elements worth noting.
Answer-first structure matters more than most people think. LLMs do not reward introductions, context paragraphs, or editorial warm-ups. They reward content that answers the question in the first sentence and builds out from there. If your lead paragraph does not contain the answer, you are probably not getting cited.
Entity specificity is the other lever. Vague content about a vague topic does not get cited. Specific content about a specific entity — a named software category, a named river, a named location, a named company — gives an LLM something to anchor its answer to. The more precisely you define the entity and its relationships, the more citable the content becomes.
Structured data helps but is not the whole story. Schema markup and structured data give LLMs clear signals about what a piece of content is and who wrote it. The JLLT citations coincided with clean schema implementation. The Bigfoot citations came from content structure more than technical markup. Both matter; neither alone is sufficient.
Topical coherence across a cluster is probably underrated. Both cases involved multiple pieces of content targeting related queries, not a single standalone article. The pattern suggests that LLMs may weight topical authority signals similarly to how Google has long weighted them for traditional ranking — a cluster of credible, consistent content on a topic is more citable than a single well-written page.
What I Do Not Know Yet
I do not have a large enough sample size to make strong causal claims. Two clients, several results, a consistent approach. That is enough to build a methodology and keep testing, not enough to write a formula.
The space is also moving fast enough that specific tactics may not transfer six months from now. What I am reasonably confident about is the underlying logic: content that is clear, entity-specific, answer-first, and topically coherent is the kind of content LLMs are built to retrieve. That is unlikely to change even as the specific mechanics do.
If you are working on this problem and want to compare notes, I am at [email protected].
Vic Shoup is an SEO, AEO and GEO strategist with 20 years of B2B organic growth experience. He is available for Director-level roles and consulting engagements. See the documented use cases →
If this was useful, you can get more like it.
Practitioner writing on SEO, AEO, and the art of working with algorithms rather than against them. One or two pieces per month.