why SEO vs. GEO misses the point

If you ask five digital marketing experts whether generative engine optimization (GEO) is different from search engine optimization (SEO), you’ll get at least eight answers.

For example:

• WIRED says to “Forget SEO,” estimating that citation overlap between search engines and AI engines has fallen to 20% (from around 70% previously).
• Partners at firm Andreessen Horowitz have declared the end of SEO in favor of a new GEO paradigm.
• On the other hand, Google representatives have repeatedly said that “good SEO is good GEO.”
• Then there’s search influencer Will Scott, who says GEO is “just SEO if you’ve been doing good SEO. But the problem is so much SEO that gets done is still the same old thing.”

So which take is right? Is there even a right take to be had?

In reality, the whole “SEO vs. GEO” conversation is a false dichotomy. 

The actual problem: Visibility has become fragmented across new and increasingly popular alternatives to traditional search. And marketers are scrambling to figure out how to adapt.

This guide will show you how visibility should be the common goal behind both SEO and GEO. 

It also provides guidance for marketing experts on how to shift focus from the Google-dominated search listings of the past to multifaceted opportunities for visibility that a more fragmented search market offers.

The fragmentation of visibility

With great optimization comes great visibility. To achieve that visibility, however, we must consider the fractured reality of the average searcher’s daily life.

Imagine someone who’s casually looking to buy a new car. They know the features and performance they want, but they haven’t quite settled on a particular make and model.

Their journey might look something like this:

• They make a few Google searches to compare models under consideration.
• Still unsure, they log into ChatGPT (or Claude or Perplexity) to ask questions that narrow down some of the options available.
• Later that evening, they scroll through TikTok when hashtag magic summons a video review of a specific model.
• They follow a link to the same creator’s YouTube channel, which launches them into a stream of long-form reviews—by the end of which, they know exactly which model they want.

Traditional SEO focuses on ranking a webpage in the search listings. But that only achieves limited visibility these days. The sheer increase in the variety of features in Google search engine results pages (SERPs)—Knowledge Panels, People Also Ask (PAA), Things to Know, Local Packs, images, videos, carousels, and so on—plus the prevalence of newer search methods, leave a lot of opportunity on the table.

Those opportunities are where the old world of simply measuring rankings and the new world of measuring visibility in a multifaceted SERP environment have yet to overlap fully. And the plethora of options available is causing some consternation.

Visibility fragmentation anxiety

Visibility fragmentation anxiety is the fear that carefully crafted SEO assets are becoming invisible as discovery shifts from traditional search engines to AI platforms. 

This fear derives from the fact that platforms offer direct answers—e.g., Google’s AI Overviews—without providing attribution, or clicks, and that the right tools do not yet exist to measure and influence their impact. 

Concurrently, ranking, as a universal metric, has been undercut by the rise of zero-click search results, as well as the emergence of new platforms that offer alternative ways to get information.

In a way, zero-click search is the realization of Google’s long-time goal of keeping users within the Google search environment (an effort that some people have called “Google Zero”). 

But in reaching this milestone, Google has had to fragment SERPs by adding features to address multiple user intents for a given query, topic, or entity.

Alternative information discovery paths

At the same time, there has been an explosion in alternative information discovery paths on other platforms, including answer-layer intermediary chatbots and social media algorithms.

AI-powered answer engines and large language models (LLMs) like ChatGPT, Claude, and Perplexity pull from a variety of sources to provide more conversational answers. 

They also let users refine and build on their questions in ways traditional search doesn’t allow. They can also help with other tasks outside the ability of Google’s search interface—such as generating images or providing workable snippets of computer code.

Likewise, as social algorithms have evolved, they’ve added greater personalization and targeting based on view history, preferences, engagement, and other factors. This has led to more people seeking advice and recommendations directly from experts, influencers, peers, and other content creators. 

Visual platforms like YouTube, TikTok and Instagram have become popular discovery destinations, especially for younger generations, while niche communities thrive on Reddit and business-oriented searchers use LinkedIn.

To address the anxiety that the fragmented visibility of these different discovery methods has created, SEOs need a new framework.

Cross-surface visibility

The good news is that we can ease this visibility fragmentation anxiety by shifting our understanding of SEO: Instead of using it as a way to achieve ranking, we should think of it as a way to achieve visibility in the places where it most matters.

In the new era of fragmented visibility, brands will need to re-strategize their optimization efforts so they appear not only in Google search result listings, but in AI-generated overviews, LLM chat responses, and various feed algorithms. Done well, a brand can use SEO to increase visibility across all the surfaces, platforms, and devices that people use to find things.

In other words, SEO should be used as a cross-surface visibility strategy, not a ranking strategy.

Historically speaking, the most visible place for brands to appear was at the top of the SERPs, especially for queries with highly relevant search intents. But that has changed, and now it’s time to change our methods along with it.

Before we can adjust, it’s important to understand how the engines behind these interfaces generate their responses.

How discovery and visibility have changed

Google’s first big algorithm change, way back in the prehistoric days of 2003, was known as the Florida update. It came right before the holidays and caused a massive disruption in how SEOs understood visibility in search results.

Twenty years later, Google launched its AI-driven Search Generative Experience, later renamed Gemini. Again, the SEO industry was forced to undergo a shift in how it thinks about visibility.

This really is just another step in Google’s mission to “organize the world’s information and make it universally accessible and useful.” Some of the optimizations the search giant has made previously include:

• Content quality: Google has consistently promoted the creation of helpful, user-focused content while demoting spammy and low-quality content. It has also introduced the concept of Experience, Expertise, Authority, and Trust (E-E-A-T) as content traits that users value.
• SERP enhancements: Google has continued to innovate and introduce better ways to present answers through features like Featured Snippets, Knowledge Panels, PAA, local packs, media carousels, and purchase and booking functionality.
• Semantic search: Shifting from keywords to topics and entities has been the impetus behind the Google Knowledge Graph, as well as encouraging the use of structured data schema.
• Technical innovation: Even before Gemini, Google used forms of machine learning and AI to add natural language processing (NLP) modules for better understanding queries (BERT and MUM) and ordering results (RankBrain).

But all of the above is related to search. What about generative AI? 

This is where the debate of SEO vs. GEO misses the point: Although LLMs like Anthropic’s Claude and OpenAI’s ChatGPT seem like completely new tools and concepts, they’re also building on the history of Google search—quite literally, at least in part.

All of the big LLMs currently in use were based, at least initially, on Google’s Transformer architecture launched in 2017. That architecture introduced methods of parallel processing and “look-back” attention that significantly improved the speed and accuracy of NLP, giving rise to the eruption of AI tools just a few years later.

While generative AI engines have all adapted transformer mechanics in different ways, they still rely on similar principles:

• Entities over keywords: They’re concerned more with understanding the relationships between entities rather than merely matching keywords.
• Context over queries: They look at the semantic context of the words used, rather than strictly answering specific queries.
• Answers over listings: They provide intelligible answers rather than simply listing content that a user has to interpret.

Let’s dive deeper into each of these principles.

Entities over keywords

Perhaps the most difficult conceptual shift for the SEO trade has been dropping the focus on keywords in favor of entities. Such a shift is necessary, however, to ensure visibility across AI-enabled platforms.

Entities represent people, places, objects, and concepts. They’re different from keywords because a keyword can represent multiple things, but entities are distinct.

For example, the keyword “orange” could represent a few different entities:

• A fruit
• A color
• The name of various locations, such as one of several counties and municipalities throughout the US, or a former principality in France

Entity-base search requires a search engine to distinguish between these entities, so it can provide users with the correct answers they’re looking for. It does so by understanding not only what entities are, but also by understanding how entities relate to each other.

Where the concept of entities clashes with historical SEO is that experience shows us using keywords has a direct, measurable impact on search rankings. As such, keyword-based SEO goes something like this:

1. Perform keyword research to find keywords to target.
2. Incorporate those keywords (and related terms) into content.
3. Monitor ranking and engagement metrics to determine the SEO value of specific keywords.

The reality is that focusing on keywords can still get a webpage ranking highly. And getting to the top of the SERPs still makes a lot of people—including a lot of C-suite executives—happy.

So if the keyword approach ain’t broke, why fix it with entities?

The problem is that the keyword approach is no longer optimal. Getting to that top spot might still be possible without an entity-based SEO strategy, but it’s not going to get your content into more visible SERP features.

Beyond Google, a keyword-based strategy is going to fail miserably. That’s because generative engines are built on a foundation of identifying entities and relating them to the questions and conversation at hand. They won’t even recognize keywords in the way traditional search engines do.

Further reading: Semantic SEO: How to optimize for meaning over keywords

Context over queries

Like the shift from keywords to entities, there’s a corresponding shift in how AI search engines understand queries and prompts. This is done with NLP, natural language processing.

Without getting too technical, NLP parses unstructured content into structured, semantically relevant components. It does that by understanding context, in particular:

• The definitions of the words used in the query
• The connotations of those words when used with each other, such as identifying sentiment 
• The typical intent of users who employ words in similar relationships to each other

Much of this context comes from the data the AI engine is trained with. However, AIs can also use personalization data that they have access to, such as demographic information, saved preferences, search or prompt history, and so forth.

When generating a response, the AI engine crafts a response based on the contextual understanding it gathered from the query.

Again like keywords and entities, the idea here is that AI search engines are doing more than simply finding keywords that match the query. 

In fact, with the lengthening of search queries into detailed, incredibly specific prompts in chat engines, it becomes less and less likely that prompts will find a match through keywords alone. Thus, relying on context becomes inevitable.

Answers over listings

A big part of the power behind AI search is the ability to provide specific answers, rather than simply linking to a webpage and implying that the answer might be there somewhere.

Google set the stage for providing context-relevant answers in 2020 with passage-based indexing. This allowed their algorithm to better understand parts of a webpage to answer specific queries, even when the overall topic of the page was not directly related. 

What began as a better way to find small bits of relevant information in a page about some other topic is now a standard part of AI Overviews.

From a visibility standpoint, this means that the main topic of a page does not limit its relevance to other topics. Likewise, content “above the fold” no longer has greater weight than the content further down the page—at least when it comes to providing an answer to a specific query. (Other factors like user experience and conversion rate optimization may still come into play.)

Practically speaking, chunking content into logical, easily digestible pieces can make it easier to receive visibility and citations in AI answer engines across the board.

Search appearance vs. systemic visibility

SEO and GEO (including AI SEO) are no longer about merely appearing in search results. They require becoming visible in ways that will show up across the entire system.

To start thinking in the right direction, it’s important to ask questions like the following:

• How often does your brand (or other entity) contribute to AI-generated responses?
• How often does your content receive linked citations in those responses?
• How often do other creators or brands mention your brand when they appear?
• What sorts of queries and prompts generate the responses that result in visibility for your brand?

When you start to track how often and in what ways your brand shows up across the board, you’ll be able to see where your efforts at achieving a more holistic visibility can be best put to use. 

Because the AI engines used by Google, LLMs, and other discovery platforms use self-learning, this goal is always going to be a moving target. But by now you know that SEO has always been a moving target.

The AI visibility measurement gap

Traditional SEO analytics track and measure progress in channel-based silos—search, social media, email, and so on. Meanwhile, newer GEO tools that monitor mentions and citations in LLMs are in the process of creating yet another analytics silo. 

To be effective in their marketing efforts, teams need to be able to see how content performs across all discovery channels, including AI search. 

AI SEO requires a new visibility framework that unifies metrics silos and covers all of the crucial ways a brand can appear:

• AI citation frequency: How often a brand appears in AI search answers over time across platforms.
• SERP rankings and impressions: Not just search listings, but all SERP features that appear for a given query.
• Social reach: Not just engagement metrics, but the brand’s full reach due to algorithmic amplification.
• Entity graph presence: The appearance (or not) alongside entities that a brand wants to appear.

So, how do we get there?

The case for unified visibility measurement

To track visibility requires measuring your brand’s footprint across as many AI search engines as possible.

Given the number of AI tools that have come out in recent years, it wouldn’t be feasible to cover them all, so we’ll look at the big ones:

• Google, including both AI Overviews that appear within search results and Gemini’s “AI Mode”
• LLMs like Claude, Perplexity, ChatGPT, and Microsoft’s Copilot 
• Social media interfaces like X’s Grok and Meta AI

As for what to track:

• Entity mentions: A simple number of times a brand or other entity appears in answers, akin to the number of keywords ranked for a term.
• Consistency of brand representation: A comparative metric showing which platforms offer better (or worse) visibility for your brand in relation to various entities.
• Engagement comparisons: How clicks stack up across platform types (search, LLM, social) and specific products.
• Branded prompts: How often people search for information about your brand compared to competitors.

Also, it’s not enough to just see snapshots. Knowing how these metrics change over time will help inform you and measure the effectiveness of AI optimization efforts.

Access to these types of metrics will help SEOs get back into the create, test, and repeat cycle that they’ve been accustomed to with Google-based ranking and engagement metrics. It’ll also help them see how optimizing content with structured data and entity-rich content can lift visibility across the board, rather than narrowing their focus to one or two platforms. 

This type of visibility tracking can also help uncover where and how discovery platforms are pulling the content that seeds their training data. This is useful for understanding how to repurpose content across delivery channels, as well as how to ensure that content uses the tools and methods developed by SEO practices.

Tools to track visibility in AIs

With the right visibility measurements, it will be possible to better understand how existing tools and methods, as well as new ones developed in the future, impact visibility across different platforms.

Some tools and methods include:

Structured data

Using schema markup to describe semantic relationships between various types of entities and a particular piece of content or webpage has been around for over a decade, and AI agents are definitely relying on that framework as well. 

Over time, it’s likely that LLMs and other AI agents will begin to support schema elements beyond Google’s structured data subset, so having a flexible framework that can publish pages with true semantic structures will be critical.

Metadata

SEOs are familiar with basic metadata elements like title tags, meta descriptions, and canonical tags. Many LLMs are paying attention to these, as well as other forms of metadata that apply to other content, such as the Open Graph protocol developed by Facebook. (And yes, Google Search has been paying attention to Open Graph for a while, now, too.)

Context control

The emergence of LLMs has given rise to a need for a new way to influence bot behavior. 

Enter llms.txt, a proposed standard for providing context and instructions to LLMs. An llms.txt doesn’t block bots from accessing content like a robots.txt file does, nor does it give a comprehensive list of website content like a sitemap. Rather an llms.txt file offers guidance by giving concise descriptions and linking to specific resources that relate to a given topic.

How SEO vs. GEO hurts teams

Another way that the SEO vs. GEO debate misses the point is in missed opportunity. Splitting efforts across different individuals and teams is often expensive and unnecessary. Instead, teams should be combining their efforts to achieve broader cross-channel visibility. 

At the end of the day, SEO and GEO share the same goals and methods. Anyone who knows how to optimize for search engines, especially from an entity-based semantic SEO perspective, already has a good grounding to optimize for generative engines—and vice versa.

Engaging in separate efforts also leads to process inefficiencies. Having one workflow to strategize, optimize, monitor, and iterate for AI SEO will avoid duplication of effort and inconsistencies in quality.

Finally, while SEO and GEO tools diverge in their coverage, it’s critical for analytical efforts to be aligned and standardized as much as possible. Such an effort becomes incredibly more difficult if the people responsible are on different teams with different budgets, priorities, and reporting structures.

At the end of the day, keeping SEO and GEO separate makes little business sense. 

Do I need to hire a GEO expert?

If you have an SEO expert on your team, chances are they already know quite a bit about GEO and AI SEO. In that sense, you probably don’t need to hire someone specifically to optimize for generative engines, since the skills involved are very similar and transferable.

However, adding new platforms to monitor and report on does increase the amount of time and attention required for that part of the job. There also may be important nuances for specific AI platforms that would be hard for a single person, or even a small team, to keep up with if they’re already stretched thin.

If you are looking to hire someone, instead of focusing on platform-specific roles, consider a skill-based division of labor. 

For example, you might consider hiring for titles like “Visibility Optimization Specialist” or “Discovery Engine Analyst” that emphasize the type of work the individual will do over the type of platform they’ll be working with.

The future of search is visibility—not SEO or GEO

Discovery engines are going to keep fragmenting the visibility landscape. But that doesn’t mean marketers need to let the SEO vs. GEO debate distract them from meeting users at the crossroads of intent and brand discovery.

To win in this new paradigm requires understanding that visibility is a cross-channel spectrum. Whether or not LLMs overtake Google as the search platform of choice, the best approach will continue to be the one that emphasizes entity-based, context-relevant, semantically focused AI SEO.

The next stage in the evolution of SEO isn’t GEO, AIO, or any other acronym ending in O (or any other letter). It’s finding a way to measure visibility across platforms and helping people discover the things they truly want to find.