From SEO to AEO: How Search Changed and Why It Matters for Your Career
In this article, you will learn: What SEO is and why it worked for 25 years. What changed when AI entered search. Five specific findings that prove the shift from SEO to AEO is real. And which parts of your existing knowledge still apply — and which don't.
This is Part 0 of the Defining AEO series on GenPicked Academy. We're building the standard definition of Answer Engine Optimization from the ground up — seven parts, grounded in evidence. Before we define AEO, we need to understand the thing it grew out of.
What is SEO, and why did it work so well?
Let's start with a quick definition. SEO stands for Search Engine Optimization. It's the practice of getting your website to appear near the top of Google's search results.
For 25 years, SEO worked because the system was simple to understand. You typed a question into Google. Google showed you a ranked list of links. You clicked one.
The whole game was: get your link ranked as high as possible. The top three results got most of the clicks. Everything below position 10 was basically invisible. Marketers built an entire industry around that feedback loop — change something on your site, watch your ranking move, repeat.
Here's the key point. The system was legible. You could see the results. You could track your position. You could measure what worked. That stability is what made SEO a career, a discipline, and a multi-billion-dollar industry.
What was the first crack?
The change didn't start with ChatGPT. It started with Google itself.
Over the past decade, Google added more and more features that answered your question right on the search page — featured snippets, knowledge panels, "People Also Ask" boxes. You've seen these. You type a question, and the answer appears at the top before any links.
Rand Fishkin, who runs a market research company called SparkToro, studied this trend using large-scale web traffic data. His 2024 analysis found that 64.82% of Google searches now end without the user clicking on anything at all. Nearly two out of three searches. The user gets what they need without leaving Google.
That was already a problem for SEO. If nobody clicks, your ranking doesn't matter as much. But the system was still Google's. The results were still mostly stable. The optimization levers — writing good content, building backlinks, fixing technical issues — still had traction.
Then generative AI made the problem structural.
How did AI change search?
When you ask ChatGPT, Claude, Gemini, or Perplexity a question, something different happens. You don't get a list of links. You get an answer — a paragraph or two that pulls together information from the AI's training data, web sources, and its own reasoning.
Think about what that means. In the old model, you saw ten links and chose which one to trust. In the new model, the AI chooses for you. It decides which brands to mention, which to leave out, and how to describe them. You read the answer. You often never see a source link at all.
That's a fundamental shift. In SEO, the goal was to earn a click. In AEO, the goal is to be inside the answer. Did the AI include your brand? Did it describe your brand accurately? Those are different questions than "did we rank on page one?" — and they require different skills to answer.
That shift — from earning a click to being part of the synthesis — is the boundary between SEO and AEO.
What still works from the SEO playbook?
Before we get to what broke in the AEO vs SEO comparison, let's be fair about what transferred. Some people will tell you everything about SEO is obsolete. That's wrong.
Good content still matters. AI models prefer clear, authoritative, well-structured content. If your content was strong enough to rank in Google, it has the raw material to be cited by AI. A 2025 Harvard Business Review article made this point well — brand authority and credibility haven't stopped mattering.
Technical basics still matter. If your website blocks web crawlers, has broken code, or loads slowly, AI systems are less likely to find and use it.
Reputation still matters. A brand with 20 years of published expertise still carries more weight than a brand-new blog. That hasn't changed.
So the SEO veterans who say "just keep doing good work" aren't entirely wrong. The timeless principles — be credible, be clear, be authoritative — still apply.
Here's what those veterans get wrong.
Five findings that prove the shift from SEO to AEO is real
1. Your Google ranking doesn't predict what AI will cite
This is the single most important finding you need to understand.
Xibeijia Guan, a researcher at Ahrefs (a major SEO analytics company), ran a study in September 2025 that tested 15,000 search queries across Google and three AI assistants. The result: only 12% of the links AI systems cited also appeared in Google's top 10 results. Eighty percent of what AI cited came from pages that don't rank in Google at all.
Let that sink in. If you're ranked number one in Google for a keyword, there's an 88% chance the AI answering the same question is citing something completely different.
Your SEO ranking is not your AEO ranking. They're different systems pulling from different sources.
2. The sources AI prefers aren't yours to control
A University of Toronto study from 2025 found that 82–89% of the sources AI systems cite are earned media. That's a term for content other people wrote about you — news articles, Reddit threads, review sites, comparison posts, blog mentions.
In the US specifically, the number was 92.1% earned content. That means the AI is overwhelmingly citing things brands didn't write and can't edit.
In SEO, you optimize your own website. You control the content, the structure, the links. In AEO, the sources that matter most are the ones you didn't create. The shift from SEO to AEO moves your optimization target from what you own to what others say about you — a pattern called earned-media bias.
That's a hard adjustment if your entire career has been in on-site optimization.
3. AI answers aren't stable
Google's search results are remarkably consistent. Search for the same thing today and tomorrow — you'll see roughly the same rankings. That consistency is what makes SEO measurement possible.
AI recommendations aren't like that at all. Rand Fishkin's team at SparkToro ran a large-scale study in January 2026 — 600 volunteers, 2,961 total prompts across ChatGPT, Claude, and Google AI. Fewer than 1 in 100 runs produced the same list of brands. Fewer than 1 in 1,000 produced the same list in the same order.
SE Ranking, another analytics firm, found it was even worse at the link level: only 9.2% overlap between three runs of the same queries on the same day.
Think about what that means for measurement. A single check of "what does ChatGPT say about our brand?" is essentially a coin flip. You'd need to ask dozens of times and look at the pattern — not the individual answer — to get a meaningful signal.
4. The meaning is stable even when the surface isn't
Here's the part that makes AEO worth doing despite the inconsistency.
A researcher at Ahrefs named Gavoyannis analyzed 730,000 query pairs comparing two of Google's AI features. The semantic similarity — how similar the meaning was — came in at 86%. But the citation overlap — which specific links they pointed to — was only 13.7%.
Translation: AI systems mostly agree on what to say. They disagree on which sources to cite when saying it. The surface churns, but the underlying meaning is stable.
This is a critical insight. If you're tracking which URLs the AI cites, you're tracking the noise. If you're tracking whether the AI understands your brand and includes it in its thinking about your category, you're closer to the real signal.
5. AI has biases that SEO didn't prepare you for
SEO had known biases — people click higher-ranked results more, familiar brands get more trust, engaging content gets rewarded. Marketers understood these and built strategies around them.
AI adds new biases you may not have heard of yet. The most important one is called sycophancy — the AI's tendency to agree with whatever the question implies. If you ask "What are the best CRMs like Salesforce?" the AI is more likely to mention Salesforce than if you just ask "What are the best CRMs?" The question shapes the answer.
There's also popularity bias — AI models over-represent brands that appeared frequently in their training data. And position bias — brands mentioned first in a long prompt get more attention than brands mentioned later.
These biases matter because most current AEO measurement tools don't account for them. We'll cover this in depth in Part 3 of this series.
Why does Google say AEO isn't real?
You might hear that AEO is unnecessary. In 2025, Google stated at a conference: "AI Mode and AI Overviews are just features of Search." Their position: standard SEO is all you need.
This is worth taking seriously. Google's AI features do use Google's search infrastructure. The technical foundations overlap.
But the data tells a different story. If AI citations matched Google rankings, Google would be right — AEO would be just SEO with a new label. The 12% overlap finding says they don't match. The 92.1% earned media finding says the source pool is different. The <1% consistency rate says the measurement challenge is different.
Google is right that good content matters. Google is wrong that good content is enough. The system changed — and understanding that change is what this series is about.
What we still don't know
A quick note on the evidence. These are strong studies, but they have real limitations.
Fishkin's consistency study used volunteer self-reporting. The University of Toronto's earned-media analysis was reported through secondary sources, not published as a direct academic paper. The Ahrefs studies focused on English-language, long-tail queries. No single study is perfect.
What makes the signal convincing isn't any one study. It's that multiple independent research teams — Ahrefs, SparkToro, the University of Toronto, Harvard Business Review, SE Ranking — all point in the same direction. The convergence is what matters.
Try this
Here's something you can do right now to see the shift from SEO to AEO with your own eyes.
- Pick a product category you know — fitness trackers, project management tools, CRM software.
- Google it: "best [category] 2026." Note the top 5 results.
- Now ask ChatGPT the same question. Then ask Claude. Then ask Perplexity.
- Compare. Which brands appear in all four? Which appear in Google but not the AI? Which appear in the AI but not Google?
The differences you see are the shift from SEO to AEO, playing out in real time on your screen.
What's next
In the next part of this series, we'll answer the question this article raised: if AEO isn't just SEO, what is it? Part 1: What AEO Is (and What It Isn't) builds the standard definition — what the field includes, what it excludes, and why the terminology is such a mess right now.
If you want to get ahead, read the AEO glossary entry for the quick definition, or explore the full AEO A to Z course for the complete learning path.
You're in the right place. Let's keep going.