GEO is the new SEO — and most brands are already losing

I searched “best CRM for mid-market companies” in ChatGPT and Claude this morning. The same three vendors came up in both. None of them were Salesforce. None of them were HubSpot.

Two of the three vendors that appeared are not the category leaders by revenue or by brand recognition. They are not the ones spending the most on Google Ads or sponsoring the most conferences. But they are the ones that LLMs have been trained to associate with credibility in that space — because of how they have structured their content, built their citation graph, and shown up in the sources that models actually learn from.

That is Generative Engine Optimization. And most marketing teams have not started.

How LLMs decide what to surface

The mistake most people make when they first encounter GEO is assuming it works like SEO — that you can optimize a page title, stuff some keywords, and climb a ranking. It does not work like that.

Language models build their understanding of your brand from the entire corpus of text they were trained on and, increasingly, from real-time retrieval from sources they trust. They are synthesizing a picture of who you are from: the quality and specificity of your public content, the sites that reference and cite you, the structured data that describes what you do, and how consistently you appear in authoritative third-party contexts.

This means the question is not “how do I rank for this keyword” — it is “what does the model believe about my brand, and what is the evidence base for that belief?”

The three ways brands appear in LLM responses

In my experience auditing how brands appear across ChatGPT, Claude, Perplexity, and Gemini, there are three modes of presence:

Cited presence: The model actively names your brand in a response and attributes a claim or capability to you. This is the highest-value placement — it carries implicit endorsement. It comes from being referenced in high-authority sources the model trusts.

Contextual presence:Your brand appears as part of a category when a user asks about the space broadly. “Companies working on X include A, B, and C.” Lower weight than a direct citation but still valuable for awareness and consideration.

Absent: The model answers the question without mentioning you at all, or — worse — describes your category and explicitly names your competitors. This is where most brands are today.

What you can actually do

Three things that move the needle faster than almost anything else:

Structured data and schema markup. JSON-LD on your site that clearly describes what you do, who you serve, and what you are known for gives models a reliable, machine-readable signal. Most brands have none. Implementing even a basic Person, Organization, and Product schema puts you ahead of the majority of your competitors.

Third-party citation building. Models trust what authoritative sources say about you more than what you say about yourself. This means editorial coverage, analyst mentions, high-quality backlinks, and presence in industry-specific repositories that models are trained to treat as credible. The content strategy goal is not ranking — it is being cited.

Specificity over volume.LLMs surface brands that are known for specific, concrete things — not brands with long lists of capabilities. “Best project management tool for remote engineering teams” returns different results than “best project management tool.” The brands that appear in the specific query are the ones that have published specific, high-quality content about that exact use case. Broad content that tries to own everything owns nothing in the answer layer.

The urgency is real

SEO took a decade to mature into a discipline that most marketing teams staffed for and budgeted around. GEO is moving faster because the adoption of AI-powered search is moving faster. The brands building presence in the answer layer right now are establishing positions that will be significantly harder to displace in 18 months.

Run the audit first. Search for your category in each of the major LLMs. See where you appear, where you don't, and who is showing up in your place. That gap is the work.