Entity SEO, LLMs, and the Fundamentals That Still Win: A Conversation with Grant Simmons

When Jeremy Rivera introduced his guest as “a legend in his own right,” Grant Simmons offered a characteristically dry correction: “I’m a legend in my own lunchtime.” That mix of self-deprecation and hard-won authority runs through the entire conversation.

Simmons describes a winding path into search. A programmer and developer back in the 1980s, he moved into marketing in New Zealand and Australia before relocating to the US in the early 1990s and joining a marketing agency. His first real SEO client was the city of Santa Monica, which wanted to rank for local hotels and restaurants. From there he immersed himself in the mid-90s LA SEO community, message boards, and early experiments, including building and ranking a video-sharing site before YouTube existed.

The heart of the conversation is entity optimisation, the area where Simmons has focused for the past several years. He frames it as clarifying the content on your site by connecting it to known information points across the web: definitive sources like Wikipedia, a drug list from a health authority, or LinkedIn as a catalogue of people and companies. Each entity tends to have a defined place in such catalogues, and the work is connecting your site’s concepts to those well-defined reference points.

His recurring image is that both Google and LLMs should be treated like small children (or “blind five-year-olds”) who need you to explicitly explain who you are, what you do, and who you serve. Hallucinations, he argues, often happen because a site has left gaps that prevent a clear, unambiguous understanding of the business.

Simmons is an ambassador for InLinks and Waikay, and he highlights a newly announced specification (entitymap.org) as an attempt to establish a standard. Rather than a plain summary of what a page is about, an entity map produces a summary with associations, connections, and disambiguation: what he calls a “meaning layer.” Deployment is still limited and the spec was announced only recently, so as he puts it, “the proof is going to be in the pudding.” But the logic tracks with what figures at Google and Bing have long described as an entity, meaning, or understanding layer.

On whether schema markup matters, Simmons stakes out a middle position. He does not believe that deeply nested schema covering every micro-concept on a page is necessary to rank better. There is a limit to the return on investment.

But he does think schema is worth using. Google leverages it for categorisation and disambiguation, Google’s AI Overviews remain the most prominent AI surface, and schema is hard to spam because you are stating in precise markup exactly what is already on the page. His rule of thumb: if it is a low lift, as with FAQ schema which most tools add automatically, there is no reason to strip it out. Even if it offers no benefit today, the expectation is that LLMs will eventually need some ontology for classification, so reinforcing meaning now is a reasonable bet.

As Rivera summed it up, much of RAG in LLMs is effectively a wrapper around Google results, so what helps Google plausibly helps LLMs too.

A through-line of the episode is that many “new” LLM-era tactics are repackaged old ones. Chunking and self-contained paragraphs? That is the passage-ranking conversation from the BERT era around 2018. Strings versus things? Simmons wrote about entity SEO back in 2013 when Hummingbird rolled out. Even the early hacks of keyword stuffing find an echo in recent stunts like white-on-white text aimed at gaming LLMs.

His point is not that nothing has changed, but that the fundamentals of information retrieval, architecture, and organisation endure; and that the industry has a habit of renaming them.

Simmons shares the story behind his long tenure at Homes.com. After the site lost roughly 60% of its organic traffic to what he diagnosed as a Panda penalty, he was hired in 2014 as its first externally recruited head of SEO. The fix required aggressive action: cutting millions of thin pages from a site of around 110 million, untangling suspicious internal linking, and taking real risks. When Panda’s rolling updates came through, the site recovered its traffic and then some. He stayed seven and a half years, rising through VP of search to VP of performance marketing before the CoStar acquisition, and went independent in 2022.

Asked what problems he is turning over, Simmons points to several:

Simmons is always after beta testers for his tools at appslicer.com. His day job is running SEO and GEO at Fiat Growth, a fintech-focused growth agency based in San Francisco, and he serves as a US ambassador for Waikay and InLinks. He is most reachable on LinkedIn.

His closing sentiment captures the spirit of the whole conversation: if one person takes away one useful thing, it was worth it; because for him, it is all about giving back.