From SEO to GEO: How to win distribution in the age of generative engines
Your next users won’t come from Google, they’ll come from AI. Here's how to make sure LLMs trust, mention, and amplify your brand.
“SEO is dead”
Says everyone on the Internet. Then why are marketers still playing that game?
The truth is, SEO is alive and well. Today, Google still processes over 14 billion searches a day, and traditional search drives 97% of queries across the web.
But your next 100k users? They’re more likely to find you through GEO (Generative Engine Optimization). Think of GEO not as a replacement, but as a branch of SEO, layering new rules on top of the old playbook.
Most companies I work with are already getting traffic from LLMs, even if they haven’t optimized for it. That’s already true for two-thirds of websites. And as generative search scales, that number will only grow.
But the challenge is that SEO taught us to optimize for keywords, while GEO demands we optimize for answers. Most teams aren’t ready for this shift.
The ones who get it right early will shape which products get discovered in the AI era.
What is GEO?
GEO is the practice of optimizing your brand, content, and authority for answer surfaces (where LLMs serve responses) instead of keyword rankings, like in SEO.
Search behavior is changing quickly. Generative engines are already processing billions of queries per month, and we’re still in the early days:
ChatGPT – 1.6B MAU
Gemini – 400M MAU
Perplexity – 20M MAU, with 780M queries/month
Claude – 19M MAU, with 88M queries/month
This massive distribution channel is also extremely high quality. A study from Semrush found that users coming through generative answers convert up to 4.4x better than those from traditional organic search.
But these engines don’t rank links, like we’re used to. They generate summarized and curated answers. When a user asks a question like “What’s the best podcast editing tool for beginners?”, the AI engine breaks it into sub-questions, pulls relevant sources, and rewrites a response mentioning the source, but not always linking back to it.
Your content is only going to surface in that response if it’s structured in a way that LLMs can parse and trust. That means human-readable, fact-rich, and widely cited content, not keyword-stuffed blog posts.
The winner is the most credible answer, not the one with the best SEO. Even if you don’t rank on Google, LLMs can still cite you, given you’ve built strong trust signals across the web.
A company I worked with, Limitless AI, is cited regularly in answers to “What’s the best productivity wearable?” They don’t rank for that search on Google, but LLMs include them because they’ve built a credible presence based on expert positioning, media coverage, and domain relevance.
What has changed
The way people search (and engines respond) is changing fast.
Here are the key changes I believe every growth team should be taking into consideration:
1. GEO provides mentions, not clicks
As a result, the percentage of clicks to sites from search engines is going down: 58.6% of Google searches now result in zero clicks. This shifts how we measure success. Instead of chasing click-through rates, we should now be measuring mentions in LLM responses.
2. Mentions happen in 4 ways
These include:
Mentions with no links
Hyperlinked text in the response
Inline citations with links
Source link carousels at the top of responses
While it’s not possible to determine exactly when AI will link directly to your website, there are types of keywords where models need to cite sources, show examples, or recommend solutions that make it more likely. For example, tool or product recommendations usually include specific brand names and links in list-style responses.
3. GEO ranks for questions, not keywords
Keywords have always been at the core of any SEO strategy. Now, GEO takes that to the next level by targeting whole groups of natural-language questions and context over keywords.
Let’s imagine we’re Descript. If we were addressing SEO, we would target podcast transactional keywords, such as “podcast editing software”. Instead, when optimizing for GEO, we would target “What’s the best tool to edit podcasts?” and other related questions.
4. Authority signals have changed
Traditional SEO rewards backlinks, domain authority, and keyword match. Now, GEO privileges citations across the web, clear and human writing that’s easy to interpret, and structured content that maps well to Q&A patterns. You’re no longer writing for a keyword parser, you’re writing for a natural language engine. And that engine is asking: does this sound like the best possible answer, written by an expert?
Understanding the changes is just the first step. Now let’s talk about what teams can do in this new reality.
How to invest in GEO
If you want to win distribution in generative engines, here are the core investments to make:
1. Prioritize your top 2 engines: ChatGPT + Gemini
All the generative engines out there will surface different results. Ask the same question, and you’ll often see results in different orders… or even entirely different answers. Trying to rank high in all of them is the fastest way to spread yourself too thin and end up with sub-optimal results across the board.
That’s why I recommend prioritizing the two LLMs with the most users: ChatGPT and Gemini. Not only do they power the bulk of traffic today, but their logic also influences many of the long-tail LLMs. It’s also where you’ll get the clearest signal. Right now, there’s very little reliable data on question volume or search behavior across LLMs. If you want to test whether your content is surfacing, start with these two.
The exception is if you have a niche product. In that case, you might benefit from focusing on a niche engine specialized in your vertical, say, an AI tool that supports academic research or developer queries.
2. Create GEO-ready content
There are a few ways to do this, and I recommend applying them all:
[A] Start with what you already know: solid technical SEO foundations, like crawlability and building strong core web vitals, such as by optimizing page speed. This ensures your content can be read by generative engines.
[B] Add answer-friendly structured content. Use tactics like FAQ schema, H1/H2 hierarchy, and clean formatting that maps well to Q&A patterns (for example, Hubspot blog posts do a great job with structured content).
[C] Identify target questions inferred via Google keywords data using tools like Ahrefs, and then proactively anticipate follow-up questions. For example, for the topic “best podcast editing software”, address questions like “what is the best tool to edit a podcast as a beginner?”, “how do I remove background noise from a podcast?”, or “what do most podcasters use to edit their shows?”
[D] Use fact-rich content backed by expert quotes, credible data, and external citations. LLMs use these references to determine whether your answer is trustworthy.
3. Build authority for LLMs
Data shows that Quora and Reddit are the most cited domains in Google AI overviews.
LLMs trust what gets cited across forums, media, and authority sites. Create visibility there, keeping in mind that authority takes time. That’s why early movers have an advantage - by the time others catch up, the engines have already learned who to trust.
OpenPhone is a prime example of this. In their early days, they did a great job by seeding discussions on Reddit, Quora, Product Hunt, and other socials across the internet. To this day, they still show consistently in AI searches within their space.
4. Consistently publish new and relevant content
LLMs value fresh and relevant content, so make sure you’re regularly publishing new GEO-ready content. Your brand’s presence in LLM answers is a moving target as responses will shift all of the time following models where retrievals vary with each query. Sometimes, getting a single forum post or backlink can tip the scale to get you a mention. To stay visible, treat your key content like live products, and ship regular updates for features and pricing as soon as they happen.
Runway does this well by consistently publishing new product updates, tutorials, and feature deep dvies throughout its site, social media, and other forums. Because of this, its content frequently appears in Perpelxity citations and ChatGPT responses.
5. Measure early success and optimize
There are tools you can use today to track early GEO performance and validate your efforts:
Track your mentions in LLM answers: tools like GPTrends and AISEO Tracker help you monitor if your brand shows up in common user questions on generative engines. Think of them as rank tracking for GEO.
Monitor traffic coming from LLMs: you can use Google Analytics 4 to identify traffic from AI-based referrals. While traffic doesn’t capture all mentions, this can give you a directional signal that your GEO efforts are working.
Benchmark your share of mentions vs. competitors: with tools like Graphite AI Tracker, you can check how often your company is being cited in generative answers and compare that to others in your space.
Taken together, these tools offer a useful early picture: are you showing up in answers or are you being left out entirely?
The companies that start building for GEO today won’t just get more traffic, they’ll shape the answers tomorrow’s users see. Authority takes time. Start early, stay consistent, and let compounding work in your favor.
Conclusion
It’s still early days, but distribution is already being highly impacted by AI.
That said, don’t believe the headlines: SEO isn’t dead. It’s still doing the heavy lifting when it comes to visibility, and it’s the foundation many LLMs are built on. Crawlability, structure, and content quality still matter.
But they’re only the first layer. You’re now supposed to be optimizing for both SEO and GEO. It’s a challenging game to play, but those who get it right and do it early will own the answer surfaces of tomorrow.
I advise growth leaders navigating the shift from SEO to GEO. If you want help building the right distribution playbook, let’s talk.