AI search is booming, but SEO is still not dead

The explosive rise of ChatGPT and other generative AI tools has fundamentally changed how we search for information online. 

These tools offer a faster, more streamlined way to access knowledge and content, often outperforming the traditional search engines we’ve come to know and rely on for decades. 

Large language models (LLMs) aren’t without flaws. But their ability to support rapid learning and problem-solving through conversational search has made a significant impact.

They’ve also opened the door to new use cases – like coding help, language learning, tutoring, and even offering a sense of companionship.

Together, these capabilities are reshaping how many people approach searching, learning, and finding information. 

For many users, LLMs increasingly feel like a more effective starting point than traditional search engines – especially when the task calls for clarity, context, or a conversational experience.

When LLMs become the starting point, what happens to SEO?

A recent study by Dr. Kevin Matthe Caramancion confirms this shift. Users generally prefer LLMs over traditional search engines for tasks requiring:

• Nuanced understanding.
• Explanation.
• Or personalized, conversational responses.

Search engines remain favored for straightforward, fact-based queries.

This change in search behavior has triggered a wave of media hype and a sense of urgency across the digital marketing industry. 

The idea that LLMs could eventually replace Google – or kill traditional search and, with it, SEO altogether – has fueled a media frenzy and a speculative gold rush, especially among those looking for an edge in AI search visibility or an opportunity to capitalize on these new trends.

New AI search tools and startups are emerging daily, fueled by the trendy narrative that traditional SEO is now outdated, and that only novel, shiny approaches like GEO (generative engine optimization) or AEO (answer engine optimization) can save the day. 

Learning to optimize for AI search is unquestionably important for business owners, marketers, and SEO professionals.

However, the narrative that this new service offering somehow implies that SEO is no longer important is some of the most inaccurate and irresponsible framing I’ve seen in my 15-plus years in the SEO industry. 

Optimizing for AI search is essential. It requires new and evolved skills, tracking and reporting, and updated knowledge, but all of this does not imply that SEO is dead.

Dig deeper: As Google shifts to AI search, legacy SEO faces a new reality

Always question where your marketing advice comes from – especially with the rise of AI

To start, it’s important to question the source of some of these big claims. 

Ironically, many of the months-old, faceless entities peddling grandiose solutions to AI search appear to be AI-generated themselves. 

They use AI logos and AI-automated marketing campaigns to offer big promises, but with little evidence that proves the “company” actually has any real experience earning visibility in AI answers.

Hundreds of self-proclaimed “GEO experts” have suddenly materialized, each professing to possess the secret key to unlocking visibility in a marketing domain that has existed for less than a couple of years. 

They claim to be experts at optimizing technology that evolves and improves daily, through reinforcement training, model updates, and new features and integrations.

This trend isn’t surprising, given the broader, pervasive hype around AI. 

That hype actively encourages the creation of AI solutions designed to replace traditional businesses, displace teams, and automate their workflows.

These sales pitches can bamboozle companies eager to jump on the AI bandwagon, without fully grasping the implications or the risks of what’s being sold.

We are, undeniably, in a cycle propelled more by hype and urgency than by genuine substance. 

Many are capitalizing on the fear, uncertainty, and doubt that have taken the marketing world by storm with the rise of AI.

This alarmist messaging has been especially effective at the executive level. 

Leadership teams are rightfully trying to understand how AI search will impact their brands and where their content fits into this new landscape. 

This raises more fundamental and existential questions for existing SEO teams.

If users begin to favor direct answers from LLMs over browsing search results, what does that mean for traditional search engine optimization?

The reality is a bit more nuanced than the alarmist messaging would have you believe.

Yes, AI search is creating a massive shift in our industry. And yes, optimizing for answer engines (AEO) requires a new set of tools, skills, and considerations.

But none of this means SEO is dead – or that the approaches SEOs have relied on for decades are suddenly irrelevant.

The opposite is true: SEO teams approaching their work the right way are at a major advantage for encouraging content and brands to appear in AI search results. 

Few professionals are more qualified to help lead clients into the future of AI search than those with experience earning visibility in search engines. 

Dig deeper: How AI is reshaping SEO: Challenges, opportunities, and brand strategies for 2025

Why SEO teams are the most qualified to help companies with AI search

As we learn more about how AI search works and how LLMs retrieve and synthesize information, it becomes clear that many of the same tactics that generate AI search visibility turn out to be common SEO approaches, only now wearing a shiny new “AI” hat.

If anyone is uniquely skilled and seasoned in navigating daily changes, it is unquestionably SEO professionals. 

Here are some of the most commonly cited tactics I’ve seen recommended as magic bullets for AI search visibility, and why they are really just standard SEO approaches that have been updated for AI search:

Focusing on conversational, long-tail searches generated during the query fan-out process

Sure, many new tools and methods exist to either directly collect and/or emulate the new query fan-out process used by LLMs, particularly Google’s AI Mode. 

Some of these tools even pull the keywords directly from Google Chrome Dev Tools during ChatGPT searches, which provide incredible insights. 

And it’s true that query fan-out uses new technology to generate relevant questions based on intent, personalization, localization, and more, as explained in great detail in this article by Mike King. 

But fundamentally, SEOs have been doing this type of query research for years. 

• Didn’t we all obsess over conversational search with the rise of voice search around 2015?
• Didn’t we start to include Speakable markup in our conversational content for this same reason, around 2018? 
• Haven’t we used long-tail keyword research generation tools, like Answer the Public around 2016, or Keyword Shitter and AlsoAsked in 2019?

The process, tools, technical considerations, and methodologies evolve (like everything in SEO).

Still, the fundamental approach to researching, conceptualizing, and optimizing for fan-out queries has been core to the work of SEOs for decades. 

Multi-modal search

One important aspect of AI search is that LLMs can process and understand information from images, videos, and audio, not just text, meaning that AI search is “multi-modal.”

But, in the words of Barry Schwartz, this is certainly “not new.” 

For years, repurposing content into different formats and earning visibility and awareness through different modes of discovery have been essential SEO approaches. 

Ross Simmonds presented on this exact topic at MozCon 2019, followed by an excellent Whiteboard Friday episode. 

SEO has been nudging us toward:

• Optimizing images.
• Creating video transcripts.
• Ensuring audio content. 

Think about how we’ve long optimized images for Google Image Search and Google Lens – or how YouTube videos have consistently appeared in standard search results.

Google began indexing podcast content in 2019, turning podcasts into a valuable source of organic discovery.

Many might argue that this work falls more into “social media” territory than SEO.

The truth is, SEO and social teams should have been working together all along. 

Our agency’s social media team uses social listening tools to help inform our SEO department about:

• What social users are saying about our clients.
• What topics are trending across different social platforms. 

Our SEO team also provides our social team with insights on:

• Which content generates the most engagement across social platforms.
• What topics are trending in search.
• Using social media to support and amplify our content for better visibility in Search and Discover. 

We both use tools like BuzzSumo and SparkToro to understand where our audiences spend most of their time and which content resonates most with them. 

Digital PR

Suddenly, “brand mentions” and “entity authority” are buzzwords for AI search, given that LLMs prioritize information from credible, well-referenced sources. 

This sounds remarkably like – digital PR. 

Looking at the most heavily cited domains and pages for nearly any brand in AI search reveals the importance of digital PR in earning AI visibility. 

For example, the following screenshot from Profound shows the most frequently cited domains across seven large language models for the brand “Canva.” 

For a company like Canva, securing positive brand mentions on trusted third-party review sites and directories is clearly vital for AI search visibility. 

Yet, a version of this truth has been fundamental to SEO at least since the inception of PageRank.

Links, mentions, co-citations, and overall brand perception have always been table stakes for healthy SEO performance.

Some SEO teams focus more heavily on digital PR and link building than others, with some partnering directly with external PR teams or agencies. 

Earning links, mentions, and visibility on trustworthy third-party websites has been a core part of the SEO conversation for decades. 

It’s also where the well-known – if sometimes cringeworthy – subcategory of “off-page” SEO first originated.

Content optimization

This one’s almost comical. 

Some of the “new” advice I’ve seen for LLM-friendly content? 

• Write clearly.
• Directly answer questions.
• Use FAQs and Q&As.
• Use structured headings.
• Break content into readable chunks and passages.
• Focus on entities.
• Provide scannable bullet points.
• Cite your sources. 

Sound familiar? 

Well, yeah. It’s an updated version of the same advice SEOs have given for effective on-page optimization for decades. 

The content that performs best in AI Overviews or chatbot responses is often precisely the content that was already optimized for:

• Featured snippets.
• People also ask boxes.
• FAQs and Q&A sections.
• General readability for human users. 

This recent commentary from SEO expert Dawn Anderson is fitting:

Cindy Krum presented about “Fraggles” at MozCon in 2019 – a term she coined by combining “fragment” and “handle.”

Her theory, which predated AI search by years, explained how Google was already indexing and leveraging “fragments” or “chunks” of content from pages and using “handles” to link directly to those specific sections. 

This idea of breaking down long pages into smaller, indexable units (passages, anyone?) to better serve user intent in formats like featured snippets is, in many ways, the foundation of how AI models now synthesize answers.

The idea that LLMs “passage-rank” or extract precise answers isn’t entirely groundbreaking or unique to generative AI.

It’s a scaled-up, evolved, and more sophisticated version of the content segmentation and retrieval Google was already pursuing years ago.

Google even confirms.

In “Your RAGs powered by Google Search technology, Part 2,” they explain how modern retrieval-augmented generation (RAG) systems, like Vertex AI Search, use technologies originally developed for Google Search.

These include breaking documents into smaller, indexable segments (passages) and re-ranking them based on relevance.

The process, involving deep re-ranking and smart content extraction, ensures LLMs receive only the most relevant, concise passages. 

It closely mirrors how Google Search has historically extracted featured snippets and indexed passages to serve precise user intents.

Note: None of this is meant to discredit or diminish the incredible work that technical SEOs like Mike King, Dan Petrovich, Andrea Volpini, Dan Hinkley, Ryan Jones, Kevin Indig, and others are doing to unpack how LLMs actually work. 

Their efforts to explore topics like passage indexing, chunking, NLP, query fan-out, Knowledge Graphs, and other core AI concepts are invaluable.

Their deep dives – through patent analysis, real-world testing, and detailed breakdowns of how AI systems interpret and synthesize information – are crucial contributions that continue to push the entire industry forward.

Cross-platform visibility

One important aspect of AI search is that large language models appear to heavily reference information found across:

• Multiple platforms.
• Data sources.
• Forums.
• User-generated content sites.

AI search doesn’t just draw from one source. 

LLMs pull in information from all over the web, including:

• Forums.
• Social media posts.
• Cloud docs.
• Random Q&A sites. 

Tools like ClickUp’s Connected Search or Microsoft’s Azure AI search are good examples. 

They scan across apps like Google Drive, Dropbox, and internal company docs to find exactly what is needed to answer the user’s query, no matter how it was phrased or where it was saved. 

AI-powered systems are also getting better at handling messy, mixed types of data (like PDF reports, Reddit threads, and blog posts), thanks to advances in natural language processing. 

LLM-powered tools act more like smart aggregators, pulling answers from across platforms instead of relying on any single site or search engine.

While this evolution has created some new implications for SEO and AI search professionals, it’s not too dissimilar from the work we’ve already done to help our clients appear in all the places their audience might discover them.

For example:

• Ecommerce SEOs have worked on optimizations for sites like Amazon, eBay, Etsy, and Walmart for years, not to mention product feed and Merchant Center optimizations. 
• News-focused SEOs have optimized Google Publisher Center, RSS, and “follow” feeds, and other news aggregator optimizations for decades. 
• Many SEOs have also driven brand awareness, traffic, and sales through TikTok, YouTube, Instagram Reels, and other video sites for many years. 

Monitoring and earning mentions, links, and a positive reputation on Reddit and other niche forums has been a core component of SEO for at least 15 years. (Earning links on Reddit and other forums was actually a big part of how I spent my first year doing SEO professionally, in 2010.)

Let’s not forget:

• The long-standing work in app store optimization (ASO) for both the Apple App Store and Google Play Store, aimed at keeping mobile apps discoverable.
• The ongoing efforts to rank in Google Images and in dedicated local directories like Yelp and TripAdvisor.

Focus on local search and NAP consistency

Large language models – particularly, Google’s AI Mode – heavily cite and reference local business information pulled from Google Maps and other business directories. 

While AI search and the shift toward Google’s AI Mode may make local SEO more important than ever, many of its core tactics have been essential for decades, such as: 

• Ensuring NAP consistency.
• Optimizing Google Business Profiles.
• Managing reviews and images.
• Other local SEO practices.

Directory submissions, anyone? It might feel like ancient SEO history, but somehow, it’s still just as relevant.

Use structured data

Many AI search-focused guidance recommends using clear and robust structured data to help LLMs better understand your content. 

While the jury is still out on the extent to which LLMs actively use structured data to understand content, there are strong arguments for its future importance, like this article by Mark van Berkel of Schema App.

That said, incorporating clear and robust structured data is by no means a new recommendation for SEO professionals. 

Using it – especially in ways that help generate rich results and boost click-through rates – has been an important part of the SEO process for years.

Still, there’s ongoing debate about how much structured data really matters for LLM and AI visibility.

Structured data expert Jono Alderson said:

• “It’s not clear if, or how, today’s (or future) AI tools use structured data – whether that’s schema.org, or something else entirely. But what we do know is that as they slurp up our HTML markup, they try to ‘make sense’ of it.”
• “So if that markup contains conveniently structured, easily tokenizable, reinforcing descriptions of what we (as marketers and business owners) believe to be the most important attributes and features of that page – in a way which also benefits other systems – then that feels like a good investment to me.”

Jarno van Driel, also a structured data expert, explained:

• “The role of on-page structured data hasn’t changed with the arrival of LLMs. It’s still very much about what it has always been about (for Search) and that is, it’s a means for search engines (mostly Google though) to enrich their results in a cost-effective way. But beyond that, on-page structured data really doesn’t achieve all that much.”
• “The presence of markup doesn’t achieve anything unless specific functionality has been created for it, so in practice this means there’s little sense in going beyond what Google mentions in its structured data documentation, whereas many seem to be convinced we should use all of schema.org to help the machines understand content.”

Dig deeper: SEO in an AI-powered world: What changed in just a year

---

Google’s own communications about AI search

Google itself has weighed in with new guidance for appearing in AI search. 

While the article mentions some of the new features and inner workings of AI search, its technical and content guidance echoes nearly all the same SEO best practices Google has shared for years.

It’s almost as if the foundations of solid information architecture and valuable content remain the same, regardless of the technological interface.

I don’t think I’m the only SEO professional who read that article and felt that its guidance looked and felt nearly identical to what Google has been saying for years.

AEO is absolutely important, but it’s incremental to good SEO, not a replacement

Now, you might be wondering whether new services like AEO are worth the investment. The answer is a resounding yes.

The main reason is that AI search demands tracking visibility, impressions, perception, and competitive share of voice across new AI-driven platforms, beyond just traditional search engines.

AEO is essential for understanding:

• How often a brand is mentioned or cited.
• What LLM responses are saying about overall brand perception, sentiment around products and services, and positioning within the competitive landscape.

That’s why it’s even more important to work with an AEO provider who can deliver these insights and reporting across multiple LLMs simultaneously, including ChatGPT, Google AI Overviews and AI Mode, and Perplexity. 

AEO services should also provide detailed, actionable recommendations on how to use these insights to boost visibility across relevant AI search platforms.

New tools like Profound, Peec AI, Otterly, WAIKAY, and ZipTie can support this process, along with updated and evolving features within major SEO platforms like Semrush and Ahrefs.

AEO requires that SEO professionals continue learning new skills, such as:

• AI and machine learning literacy.
• Prompt engineering.
• Vector embeddings and cosine similarity analysis.
• NLP and semantic search.
• Workflow automation.

AI search hasn’t made a dent in Google Search – yet

Despite all the hype around AI search, each new study suggests it has yet to make a meaningful dent in Google’s dominance. 

Google processed an astounding 14 billion searches per day in 2024, per Rand Fishkin of SparkToro. This is 373 times more than the estimated daily “search-like” prompts on ChatGPT. 

Even if all of ChatGPT’s 1 billion daily messages were search-related, its market share would still be under 1%, while Google holds over 93.57%.

Google’s search volume actually grew by more than 21% in 2024 compared to 2023, based on data from Datos. 

While some users may be turning to large language models like ChatGPT to start their queries, it’s clear that Google hasn’t been abandoned, at least not yet.

This perspective is echoed in Glenn Gabe’s recent analysis, which found that AI search was driving less than 1% of traffic to most websites as of June 2025, with many seeing numbers below 0.5%.

Gabe cautions against an overemphasis on AI search at the expense of traditional Google SEO. 

Neglecting core search quality can lead to negative consequences from Google’s broad core updates. 

For example, churning out lots of low-quality content in an attempt to gain AI search visibility can cause a site to see declines in visibility with future core updates.

And because Google’s AI Mode and AI Overviews draw from Google’s own search index (see the below image from Gabe’s article), core updates can impact a site’s visibility across all search features, including AI search platforms. 

It’s not just Google’s AI products that rely on search results. 

Search results play a major role in how other large language models, like ChatGPT, Perplexity, and Claude, generate answers for search-enabled queries.

Dig deeper: How to stay grounded (and inspired) as AI changes search and SEO

AI search draws from traditional search signals and may even leverage Google’s search results

One of the most impactful ways AI search builds on existing SEO performance is through retrieval-augmented generation. 

RAG enables language models to incorporate real-time, external information, rather than relying solely on pre-trained data, which can quickly become outdated. 

When an LLM identifies that current web content could improve a response, it acts as a “retriever,” pulling relevant information from sources like Google or Bing search results.

It’s long been clear that Google uses its own search results for RAG. 

However, there’s growing speculation that ChatGPT may now be doing the same, pulling from Google results, not just Bing’s (which OpenAI has previously confirmed in its documentation).

Alexis Rylko‘s recent article speculates that OpenAI may have quietly switched from pulling in Bing’s results to strongly leveraging Google’s own search results in its responses. 

Some of Alexis’ findings mirror my own discovery from four months ago – specifically, ChatGPT citing URLs that include Google’s unique ?srsltid parameter.

Similarly, my colleague Johnny Herge recently found examples of ChatGPT citing Google Maps URLs in its recommendations for local businesses. 

This convergence of AI search and traditional SEO is perhaps the most compelling argument for SEO’s essential role in the age of AI.

If LLMs are actively retrieving and synthesizing information from existing search results, particularly Google’s search results (the search engine most SEOs have focused on for 20+ years), then the content that ranks prominently in traditional Google Search becomes the de facto source material for the LLMs that rely on it.

This also reinforces my belief that many of the newer spammy techniques aimed at gaining AI search visibility are dangerous. 

Tactics like adding prompt information using white text on a white background may offer a temporary boost, but they’re essentially recycled 20-year-old SEO spam strategies. 

Just because something works in AI search today doesn’t mean it will continue to as LLMs evolve and release improved models.

We are in an AI-driven hype cycle, and it’s not the first hype cycle in SEO history

It’s important to pay attention to how AI is impacting search and how users are increasingly using it. 

That said, anyone working in SEO for more than a decade will recognize this current “AI apocalypse” narrative as a bit of an SEO throwback, echoing countless past declarations of SEO’s impending doom.

(Spoiler alert: SEO wasn’t doomed; it’s more alive now than ever.)

Below are some other examples of technological developments that created significant hype about their existential threat to search over the years.

Google+ and Facebook (early 2010s)

The early 2010s brought an industry-wide obsession with “social signals,” fueled by the rise of platforms like Google+ (launched in 2011) and the dominance of Facebook. 

Many believed that social engagement would soon render SEO irrelevant.

Despite the hype, Google+ ultimately failed to gain significant traction.

While social media remains vital for branding, its direct impact on organic search usage was consistently shown to be minimal, never fully replacing traditional SEO.

Mobile-first indexing (mid-2010s, officially rolled out 2018)

Mobile-first indexing created a panicked frenzy to adapt to Google’s shift to prioritizing mobile versions of websites for ranking, with dire warnings of being “left behind” or “invisible” if not immediately compliant. 

The buzz began years before the official rollout in 2018. 

SEO teams evolved their services to ensure client websites were compliant with mobile-first indexing and retained visibility in an increasingly mobile world.

It ended up being much more gradual than expected, as many major brands took years to comply with the new guidelines, without losing much visibility in the meantime. 

In retrospect, mobile-first indexing represented a needed evolution that solidified best practices rather than creating a sudden, life-or-death ranking factor. 

The real impact was much more nuanced and gradual than what was originally expected.

Voice search (mid-to-late 2010s)

This era of SEO was marked by the widespread belief that spoken queries, amplified by voice search devices like Amazon Echo (2014) and Google Home (2016), would soon replace typed searches altogether. 

Many predicted that keyword research would become obsolete and content strategies would need to shift entirely to suit conversational AI.

Voice search, however, ended up being a bit of a dud, even evolving into a recurring inside joke within the SEO community. 

There was little evidence that consumers actually replaced typed searches with voice, as typed queries continued to grow exponentially. 

And while the popular statistic claimed that by 2020, 50% of all searches would be done by voice, that prediction still hasn’t come to pass.

Page speed and Core Web Vitals (late 2010s to early 2020s)

The intense focus on technical performance metrics, particularly with the introduction of Core Web Vitals in 2020-2021, sometimes to the exclusion of content quality and user experience, drove enormous fear of algorithmic penalties.

Core Web Vitals, while important, were later felt to be a bit overblown, especially as they came to be known as a “tiebreaker” signal, not the earth-shattering ranking factor many feared would tank the rankings of non-compliant websites.

TikTok and video (early 2020s)

The pronouncements that short-form video, largely popularized by TikTok, would render traditional text-based search irrelevant caused panic at Google and caused many to suggest that TikTok posed an existential threat to Google’s search business. 

But not only did Google respond with its own short-form video product, YouTube Shorts. TikTok’s rise also hasn’t significantly disrupted Google’s dominance in search.

Dig deeper: 6 easy ways to adapt your SEO strategy for stronger AI visibility

The unshakeable core of SEO: Sorry, but SEO is still not dead (yet)

Another year, another declaration that SEO is dead. But the truth reveals another story.

It’s clear that AI search is a powerful, evolving force that should not be ignored. 

The rise of LLMs demands new skills, new metrics, and at the very least, an insatiable desire to continue learning and experimenting.

Services like AEO are genuinely important for understanding brand visibility and perception within these new AI platforms. 

And yes, for businesses, being at the forefront of optimizing for these emergent platforms is not only advisable but essential.

However, the narrative of traditional SEO’s demise is not just premature; it’s profoundly inaccurate. 

As we’ve explored, the very mechanics of AI search – from retrieval-augmented generation pulling from existing search results to its reliance on established signals like clear content optimization, E-E-A-T, structured data, and cross-platform authority – underscore the enduring relevance of traditional SEO approaches. 

Also, the content that performs best in AI Overviews or receives citations from ChatGPT is, more often than not, the same high-quality, well-optimized content that excels in traditional Google Search.

This isn’t an existential threat; it’s an evolution. 

And if any profession is uniquely skilled and seasoned in navigating daily evolution, it is unquestionably SEO professionals. 

For decades, we’ve adapted to algorithm updates, new search features, shifting user behavior, evolving spam and content policies, and a steady stream of so-called “SEO killers” – none of which have actually succeeded.

SEOs have learned to optimize for mobile, voice, video, images, local searches, and app stores, mastering a myriad of platforms beyond just websites. 

AI search is simply the next frontier for SEO professionals to conquer, much like the many previous search iterations before it.