a16z: GEO Will Rewrite the Rules of Search

01

Foreword

Today, let’s read an article together — “How Generative Engine Optimization (GEO) Rewrites the Rules of Search”, published by a16z on May 28, 2025.

In keeping with the spirit of immersive reading, I’m simply excerpting the article to make it easier to read, without adding any personal interpretation.

That said, I plan to publish my own analysis of GEO-related content in the near future. If you’re interested in this topic, stay tuned. I’ll be dedicating more attention to GEO going forward.

Let’s dive in. The article is quite long, so feel free to read it in parts or skip to sections you find most interesting.

02

GEO is Rewriting the Rules of Search

“It’s the end of search as we know it, and marketers feel fine. Sort of.”

The era of search as we’ve known it is coming to an end — and marketers are… surprisingly calm. Sort of.

For over 20 years, SEO (Search Engine Optimization) was the go-to strategy for online visibility. It gave rise to an entire ecosystem of keyword stuffers, backlink traders, content optimizers, auditing tools — and the professionals and agencies behind them. But in 2025, search is shifting away from traditional browsers toward LLM (Large Language Model) platforms. With Apple’s announcement that AI-native search engines like Perplexity and Claude will be built into Safari, Google’s distribution dominance is being questioned. The foundation of the $80+ billion SEO market is starting to crack.

A new paradigm is emerging — one not powered by page rank, but by language models. Welcome to Act II of search: Generative Engine Optimization (GEO).

03

From Links to Language Models

Traditional search was built on links. GEO is built on language.

In the SEO era, visibility meant getting a top spot on the results page — determined by keyword matches, content depth, backlink volume, user engagement, and more. Today, with LLMs like GPT-4o, Gemini, and Claude becoming the primary interface for how we find information, visibility means appearing directly in the answer, not just ranking high in a list of blue links.

As answer formats evolve, so does the way we search. AI-native search is becoming fragmented across platforms like Instagram, Amazon, and Siri — each driven by different models and user intents. Queries are longer (23 words on average vs. just 4), sessions last longer (about 6 minutes), and responses vary depending on context and source. Unlike traditional search engines, LLMs remember, reason, and synthesize answers from multiple sources, tailored to the user. This fundamentally changes how content is discovered — and how it should be optimized.

Traditional SEO rewarded precision and repetition.
Generative engines favor content that is well-organized, easy to parse, and semantically rich — not just stuffed with keywords.

Helpful phrases like "in summary" or bullet points can improve content’s extractability for LLMs.

It’s also important to recognize that LLMs have a very different business model than traditional search. Search engines like Google monetize through ads — users pay with data and attention. In contrast, most LLMs are subscription-based and sit behind paywalls. This structural shift affects content referencing: model providers are less incentivized to surface third-party links unless it improves user experience or product value.

While we may eventually see ad ecosystems built on top of LLM interfaces, the rules, incentives, and players will look very different from traditional search.

In the meantime, outbound clicks are becoming an emerging indicator of value. ChatGPT, for instance, is already generating referral traffic to tens of thousands of domains.

04

From Page Rank to Model Relevance

It’s no longer just about click-through rate — it’s about reference rate: how often your brand or content is cited in model-generated answers.

In a world dominated by AI outputs, GEO is about optimizing to be referenced, not just to show up on a list. That shift is redefining how we measure brand visibility and performance.

New platforms like Profound, Goodie, and Daydream are already enabling brands to monitor how they’re appearing in AI responses, track sentiment, and understand which publishers influence model behavior.

These tools fine-tune models using brand-specific prompts, inject high-value SEO keywords, and run synthetic queries at scale. Their outputs are consolidated into dashboards that help marketing teams track visibility, ensure consistent messaging, and analyze competitive positioning.

Case in point: Canada Goose used such tools to understand not just whether its products (e.g., warmth, waterproofing) were being referenced — but whether the brand itself was spontaneously mentioned by LLMs. In the AI era, unaided brand recall by a model is a key metric.

These monitoring tools are becoming as essential as traditional SEO dashboards. Ahrefs’ Brand Radar, for example, tracks brand mentions in AI Overviews, while Semrush has released an AI toolkit to help brands manage perception, optimize visibility, and respond in real time to emerging mentions.

We’re witnessing a new type of brand strategy: not just managing public perception, but managing model perception.

Your presence in the AI layer is the new competitive advantage.

Of course, GEO is still experimental — much like SEO was in its early days. Every major LLM update can require relearning how best to optimize. Just as Google’s algorithm updates disrupted SEO in the past, LLM providers are constantly tweaking citation logic.

There are already multiple schools of thought around GEO. Some tactics are clear (e.g., being cited in source documents), while others are more speculative — such as whether models prefer journalism over social media, or how different training sets influence preferences.

05

Lessons from the SEO Era

Despite its scale, SEO never produced a true monopoly. Tools like Semrush, Ahrefs, Moz, and Similarweb succeeded, but each dominated a niche — backlinks, traffic, keywords, or audits — and none fully controlled the full stack.

SEO remained fragmented, distributed among agencies, internal teams, and freelancers. Rankings were inferred, data was messy, and Google controlled the algorithm — but no one controlled the interface.

Even at its peak, SEO tools lacked the user stickiness or network effects to become true hubs. One key limitation was access to clickstream data — the clearest window into real user behavior — which was locked behind ISPs, SDKs, extensions, or brokers. Without privileged access, building accurate, scalable insights was nearly impossible.

GEO changes that.

06

How to Get Referenced: The Rise of GEO Tools

This isn’t just a tooling shift — it’s a platform opportunity.

The most powerful GEO companies won’t just measure — they’ll fine-tune their own models, learning from billions of implicit prompts across sectors. They’ll own the feedback loop: insight, creative input, feedback, iteration.

They won’t just observe LLM behavior — they’ll influence it.

They’ll also find ways to tap into clickstream data and integrate first-party and third-party sources.

Winning GEO platforms will provide actionable infrastructure: generate real-time campaigns, optimize for model memory, and iterate daily as LLM behavior evolves.

These systems won’t just monitor — they’ll operate.

And the opportunity goes far beyond visibility. GEO isn’t just how a brand gets referenced — it’s how it builds an ongoing relationship with the AI layer.

GEO becomes the system of record for interacting with LLMs — tracking performance and presence across platforms. Control that layer, and you control the budget behind it.

That’s the monopolistic potential: not just providing insights — becoming the channel.

If SEO was decentralized and data-adjacent, GEO is centralized, API-driven, and embedded directly into brand workflows.

While GEO might seem like just the next evolution of search, it’s actually a wedge into a much broader space: performance marketing. The same brand guidelines and user data that drive GEO also power growth strategies.

This is how big businesses are built — by testing and optimizing across multiple channels.
AI enables the rise of the autonomous marketer.

Timing matters. The shift in search has only just begun, but ad dollars move fast — especially when there’s arbitrage.

In the 2000s, that arbitrage was Google Adwords. In the 2010s, it was Facebook’s targeting engine. In 2025, it’s LLMs — and the platforms helping brands manage how their content is consumed and cited by these models.

Put simply, GEO is a battle for space in the model’s mind.

In a world where AI is the front door to commerce and discovery, marketers must ask:
Will the model remember you?