The Search Engines You're Not Optimizing For

Why your brand might be invisible to the fastest-growing audience in marketing

Eight percent of consumers are now choosing ChatGPT over Google for search. That number was one percent six months ago. If you think that’s not your problem yet, you’re already behind.

The conversation about large language models in marketing has focused mostly on content creation and customer service chatbots. What’s being overlooked is that LLMs are becoming a discovery layer between consumers and brands. And most brands aren’t even showing up.

How search is actually changing

When someone asks ChatGPT “what’s the best project management software for a small marketing team,” they don’t get a list of ten blue links to click through. They get a direct answer with specific recommendations. Maybe Asana gets mentioned. Maybe ClickUp. Maybe your product doesn’t get mentioned at all.

That’s the problem. Traditional SEO assumed people would see your link in search results and click through. LLM search assumes the AI will answer the question directly, and the conversation is over. If you’re not in that answer, you lost the opportunity entirely.

The citation economy

Some LLMs cite their sources. Many don’t. ChatGPT with browsing enabled will sometimes include links. But users aren’t necessarily clicking them. They’re getting their answer from the AI’s summary, not from your carefully optimized landing page.

This creates a new dynamic. It’s not about ranking #1 anymore. It’s about being mentioned at all. It’s about the LLM having enough information about your brand to include you in relevant answers. And right now, most companies have no idea whether they’re being mentioned or not.

Perplexity has started testing “sponsored follow-up questions” where brands can pay to have their products surfaced first in certain categories. That’s the beginning of LLM advertising. It won’t be the end.

The training data advantage

Here’s the uncomfortable part: the brands winning in LLM search are often the ones who were winning in traditional search. Google’s training data for Gemini includes billions of web pages. OpenAI’s training data for GPT included massive amounts of web content. If your brand had strong SEO, good content marketing, and lots of mentions across the web, you probably show up well in LLM results. If you didn’t, you probably don’t.

But there’s a twist. Recency matters differently. LLMs often have training cutoffs, so recent news and recent content might not be reflected. A brand that launched six months ago might barely exist in the model’s knowledge. A brand that had huge SEO presence three years ago might be over-represented relative to their current market position.

Share of model vs. share of voice

There’s emerging research on something called “Share of Model” - essentially, what percentage of relevant queries result in your brand being mentioned by the LLM. It’s like share of voice in traditional advertising, but for AI recommendations.

Early data suggests massive disparities. In some categories, the top three brands mentioned by LLMs capture 80% of all recommendations. The rest get nothing. Traditional search at least gave you a chance to show up on page one if you had decent SEO. LLM search is more binary: you’re either in the answer or you’re not.

What GEO actually means

Generative Engine Optimization is the new term being thrown around for LLM optimization. Companies like Evertune and Daydream are building businesses around helping brands show up better in AI search results.

But what does that actually mean in practice? It’s not about keyword stuffing or backlinks. It’s about having structured, authoritative, comprehensive information about your brand easily accessible online. It’s about being mentioned in context by trusted sources. It’s about ensuring your key category associations are well-documented.

Think of it less like traditional SEO and more like brand positioning that happens to be machine-readable.

The retail media parallel

This shift feels similar to what happened with retail media networks. Amazon became the starting point for product search. If you weren’t showing up well on Amazon - either organically or through advertising - you were missing a huge chunk of purchase consideration. Traditional SEO on Google became less relevant for product discovery.

LLMs are creating a similar dynamic, but broader. They’re not just for product search. They’re for recommendations, advice, research, and general information. If someone asks an LLM about skincare routines, meal planning, productivity tools, or vacation destinations, brands either get mentioned or they don’t.

The human-AI awareness gap

Here’s what keeps me up at night: there’s a gap between how consumers perceive brands and how AI models represent brands. A consumer might have had a great experience with your product, but if that sentiment isn’t captured in training data, the LLM doesn’t know about it.

Conversely, negative press or criticism might be over-represented in training data, causing the LLM to have a more negative view of your brand than actual consumers do. The model’s perspective becomes a distorted mirror of reality.

What nobody’s saying about LLM advertising

When Perplexity and ChatGPT start selling advertising at scale - and they will - it’s going to be expensive. These platforms have massive user bases but limited ad inventory compared to traditional search. They don’t have ten ad slots per results page. They have maybe one or two moments to insert a commercial message.

The early advertisers in these channels will pay a premium. But they’ll also get to define what LLM advertising looks like. Just like Google AdWords early adopters got incredible CPCs and learned the playbook before it became competitive.

The strategy you actually need

Stop waiting for someone to write the definitive guide to LLM optimization. Start experimenting now. Here’s what that looks like:

First, audit where you currently show up. Ask ChatGPT, Claude, Gemini, and Perplexity about your category. See if your brand gets mentioned. See what they say about you.

Second, ensure your foundational content is accessible and structured. LLMs need clear, authoritative information. Your website should have well-structured pages about your products, your company, your key differentiators. Not marketing fluff - actual substantive content.

Third, increase your presence in trustworthy sources. Get mentioned in industry publications, case studies, research reports, and credible reviews. These sources are more likely to be weighted heavily in LLM training data.

Fourth, monitor the conversation. Use tools that track LLM mentions. Track how often you’re being recommended. Track the context. Track compared to competitors.

The timing problem

The challenge is that by the time definitive best practices emerge for LLM optimization, the advantage will be gone. Early movers will have established strong presence in training data. They’ll be the default recommendations. Changing that will be harder than building initial presence.

Traditional SEO took years to master, but it also took years for results to compound. LLM optimization might happen faster because these systems are being retrained more frequently. But that also means the window to establish presence is shorter.

What this means for smaller brands

If you’re a smaller company without the resources for massive content marketing or widespread press coverage, LLM search creates both opportunities and challenges.

The opportunity: LLMs don’t necessarily favor big brands the way Google’s algorithm does. If you have genuinely better products or more innovative solutions, and that’s documented somewhere credible, you might get recommended alongside bigger competitors.

The challenge: you need someone to write about you. You need to be mentioned in contexts that end up in training data. That requires relationships, PR, and content that others want to reference.

Traditional SEO let small companies compete through technical excellence and smart content. LLM optimization might require more external validation and third-party endorsement.

The future of discovery

Search isn’t going away. But search behavior is changing. Younger users especially are starting product research by asking AI rather than typing keywords into Google. They’re having conversations with LLMs rather than clicking through multiple websites.

This doesn’t mean traditional SEO is dead. It means it’s becoming one piece of a larger discovery ecosystem. You need to show up in Google. You also need to show up in ChatGPT. And Perplexity. And whatever comes next.

The brands that figure this out early - that build systematic approaches to LLM optimization, that track their Share of Model, that secure early advertising positions - will have an advantage that compounds over time.

The ones waiting for the dust to settle might find that by the time they act, consumer behavior has already shifted, and they’re playing catch-up in a channel where established presence is hard to displace.