Every few years, someone declares that SEO is dead. It happened when Google launched Panda. Then Penguin. Then RankBrain. Then featured snippets. Now it’s AI Overviews and ChatGPT.
And every time, the obituary turns out to be wrong.
What’s actually happening isn’t the death of SEO. It’s more a refactoring of what SEO means. The problem is that the people running most businesses haven’t updated their definition since around 2009.
What Business Leaders Think SEO Is
Ask most executives to describe SEO and you’ll get some version of the same answer: keywords, rankings, maybe backlinks. A number on a dashboard. Something you pay an agency to do, something that probably involves some amount of wizardry, and something that should result in your website showing up first on Google.
That’s not entirely wrong. It’s just about fifteen years out of date.
The SEO industry spent most of the 2000s and early 2010s playing a game of “trick the algorithm.” Stuff enough keywords into a page, build enough (often low quality) backlinks, and watch the rankings climb. It worked, until it didn’t. Google got smarter. The tricks stopped working. The agencies that survived were the ones that had been doing it right all along.
What they were doing right looked less like hacking a system and more like building genuine authority: creating content that actually answered real questions, earning trust from credible sources, making their clients’ websites fast and accessible and structured in a way that was easy to understand. Less manipulation, more SEO as promoting expertise.
That shift happened gradually, then all at once. But a lot of boardroom definitions of SEO never caught up.
Now Add AI to the Conversation
When AI-generated search results started appearing, (e.g., Google’s AI Overviews, answers surfaced directly from ChatGPT and Perplexity) the “SEO is dead” crowd got louder. And honestly, you can understand why. If people are getting answers directly from AI without ever clicking through to a website, what’s the point of ranking?
Here’s what that argument misses: the AI has to get its information from somewhere.
When an AI tool summarizes an answer to a question, it’s drawing on content it has determined to be credible, authoritative, and well-structured. It’s citing, or at least drawing from, sources it trusts. The criteria it uses to make that determination look a lot like the criteria a search engine uses to rank content.
Topical authority. Clear and well-organized information. Demonstrated expertise. A track record of being referenced by others. These are the things that make content rankable in traditional search. They are also the things that make content usable and citable in AI-generated responses.
The Web Is Still How AI Learns About the World
To understand why web presence still matters, it helps to understand how large language models actually get trained.
Most major LLMs, the models powering ChatGPT, Google Gemini, Meta’s LLaMA, and others, are trained on massive datasets built primarily from web crawls. The largest and most widely used of these is Common Crawl, a nonprofit that has been systematically crawling and archiving the public web since 2008. Its dataset currently covers billions of web pages and petabytes of text. It is, quite literally, a snapshot of what the internet knows and it forms the backbone of training data for most of the AI tools your customers are using right now.
But Common Crawl doesn’t take everything. The raw crawl gets filtered and curated before it ever reaches a model. Researchers and engineers apply quality signals to decide what makes the cut: pages with meaningful inbound links, content from domains with established authority, well-structured text that demonstrates depth and coherence. Low-quality pages, thin content, and sites with no credible external references tend to get filtered out. The criteria aren’t identical to Google’s ranking algorithm, but the underlying logic that authority, structure, and genuine usefulness are signals of quality is strikingly similar.
For AI tools that operate in real time, Perplexity, ChatGPT with web search enabled, Google’s AI Overviews, the relationship is even more direct. These systems crawl the live web, evaluate pages using quality signals that closely mirror traditional SEO metrics, and use that evaluation to decide whose content gets cited in the answer. They are, in effect, running an accelerated version of a search engine’s ranking process every time someone asks a question.
Here’s the point that often gets lost in the “AI killed SEO” narrative: the web remains the medium with the lowest barrier for inclusion in these systems. A proprietary database isn’t crawlable. A PDF sitting on a file server isn’t indexed. A LinkedIn post disappears into a walled garden. But a well-structured, authoritative, publicly accessible web page? That’s exactly what Common Crawl is built to find, what Google is built to rank, and what AI search tools are built to surface.
If you want your business to exist in the answers AI generates for your potential clients, the path there runs directly through your web presence — and through the same qualities that have always determined whether a web presence is worth anything.
The Fundamentals Don’t Change. The Surface Does.
The agencies that will survive this transition, and help their clients thrive in it, are the same ones that survived every previous transition… the ones built on fundamentals rather than tricks.
What does that look like in practice?
It means creating content with real depth, written by people who actually understand the subject, structured in a way that’s easy for both humans and machines to parse. It means building genuine credibility — the kind that comes from being referenced, linked to, and cited because you’ve earned it. It means making sure the technical foundation of a digital presence is clean and accessible. It means understanding not just what your audience is searching for, but what questions they’re trying to answer and what would actually help them.
None of that is new. It’s what good SEO has always been. It’s also, not coincidentally, exactly what makes content surface in AI-generated responses.
The surface has changed. The search bar looks different. The result might be a paragraph instead of a list of blue links. But the underlying logic — what earns authority, what builds trust, what gets cited — is remarkably consistent.
What This Means If You’re Running a Business
If your current SEO strategy is built around ranking for a list of keywords, it’s not wrong — but it’s incomplete. The question to be asking is broader: Where does my audience go when they want answers, and am I present in those places?
That includes traditional search. It includes AI-assisted search. It includes the tools your potential clients are increasingly using to do preliminary research before they ever talk to a vendor.
Showing up in all of those places requires the same thing: being genuinely authoritative on the topics that matter to your audience, and making that authority easy to recognize and easy to reference.
The businesses that treat this moment as a reason to abandon their SEO investment are making a mistake. The ones that treat it as a reason to double down on the fundamentals — real expertise, real content, real credibility — are going to be very well positioned.
SEO isn’t dead. It just finally means what it should have always meant.