Why AI Search (GEO/AEO) Is Eating Traditional SEO, and What Agencies Must Do Now
If you're still measuring SEO success by ranking positions, you're measuring the wrong thing.
I'll be honest: that sentence used to feel a little dramatic. It doesn't anymore. In 2024, Google's AI Overviews appeared on more than half of search queries in some categories. Perplexity went from niche to mainstream. ChatGPT became the first stop for millions of people who used to Google things. And a growing chunk of that traffic just stopped clicking through to any website at all.
Old-school SEO assumed one behavior. Person types query. Ten blue links appear. Somebody clicks.
That model is breaking in plain sight. Not everywhere, not overnight, but fast enough that agencies are seeing odd client dashboards: rankings steady, impressions weird, clicks down, conversions harder to explain. Why does this matter? Because the old reporting story no longer matches how people actually search.
Here's what's happening, why it matters, and three tactics we've found useful.
What GEO and AEO actually mean (and why the distinction matters)
Generative Engine Optimization (GEO) is writing and structuring content so AI systems (ChatGPT, Perplexity, Google AI Overviews, Claude, Gemini) will surface, cite, or summarize it when answering a query. Unlike traditional SEO, GEO isn't really about links or domain authority. It's about whether your content is usable by an AI as a source.
Answer Engine Optimization (AEO) is the older sibling. It focuses on getting content into featured snippets, voice search results, and structured answer boxes. Those were the precursors to today's AI answers. AEO taught one useful lesson: format matters. Questions help. Direct answers help. Lists help, but not every answer needs to be chopped into a neat little list.
Most guides blur GEO and AEO together. That's only half right. AEO was about winning answer surfaces inside search engines. GEO is about becoming source material for systems that may never show a classic search results page at all.
Why the distinction matters: you can rank #1 organically and still get zero visibility in AI answers. We've seen it. A client sits at position one for a moderately competitive keyword, but when users ask ChatGPT or Perplexity the same thing, our client's site doesn't show up at all. A competitor with weaker backlinks but better-structured content does.
That stings.
What the data is actually showing
I want to be careful here and not invent numbers. What we see across client accounts is directional but real. In categories where AI Overviews appear often (health, finance, how-to, product comparisons), organic CTR has dropped 15–35% compared to 2022 baselines for the same rankings. Some queries that used to drive 400 clicks a month now drive 80.
A 2024 BrightEdge study found AI Overviews appeared on over 50% of queries across several verticals. Semrush data from late 2024 showed that position-one results in AI-heavy SERPs were getting less than 2% CTR in some categories.
The shift isn't uniform. Transactional queries with local intent are still relatively safe for traditional SEO. So are branded queries. Queries with strong visual intent have held up better too. The damage is concentrated on informational queries, especially the "what is," "how to," and product-comparison pages content teams have been publishing for years.
Counter to the usual panic, this does not mean every blog is dead. It means lazy informational content is a weaker asset than it used to be.
That's a lot of content investment quietly losing force.
Three tactics that actually work for GEO
I'll skip the vague advice. Here's what we've actually tested.
1. Write definitions AI systems can quote verbatim
AI systems want dense, self-contained answers. If your page defines a concept, that definition should work as a standalone 30–50 word block that answers the question without any surrounding context. Not "as we discussed above, this approach is effective because..." but a clean, quote-ready statement.
We rewrote several FAQ sections to follow this pattern. State the term. Define it in one tight sentence. Add one specific supporting fact or number when you have one. Pages that went through this treatment started appearing in Perplexity answers within 6–8 weeks.
Correlation, not proven causation. My take: that caveat matters, but it doesn't make the pattern useless. We've seen it often enough now to keep doing it.
2. Answer before you explain
Traditional content structure often buries the answer. Introduction, context, background, then answer. AI systems scrape for the direct response, and if it's on page three of your long-form article, they might find it or they might not.
The shift is simple: lead every section with the answer. One sentence, direct, no hedging. Then follow with explanation, examples, caveats, source context. This is sometimes called BLUF structure (Bottom Line Up Front), borrowed from military writing.
Is this overkill? For a 50-page site, no. It makes content more likely to be lifted into AI answers, and it also just makes for better writing.
3. Use named experts and specific sources, not "industry experts say"
Vague attribution ("experts believe," "studies show") is almost invisible to AI citation algorithms. Named sources with dates and publication context are not. When you write "according to a 2024 BrightEdge survey of 1,700 enterprise marketers," you're giving AI systems something to verify and attribute. When you write "experts agree," you're giving them nothing.
This is more work. No way around it. It means actually citing sources instead of gesturing at them.
Yes, this contradicts some old SEO advice about keeping pages clean and frictionless. Bear with me. For GEO, attribution is not clutter; it's a parsing aid. In our experience, pages with specific attributions consistently outperform vague ones in AI citation surfaces.
What definitely doesn't work anymore
Keyword density. If you're still manually targeting a keyword 12 times per 1,000 words, you're doing something that no longer correlates with AI visibility. AI systems understand topic coverage, not keyword repetition.
Skip this step.
Thin "pillar pages" with no original insight. A 3,000-word page that summarizes what five other pages already say is not a GEO asset. AI systems tend to prefer sources that add something: a specific dataset, a case study, a defined methodology, an original opinion. Generic overviews are easy to skip when there are thousands of them.
Ignoring structured data for Q&A content. FAQ schema, how-to schema, and article schema don't guarantee AI citation, but they help AI systems parse your content structure. Skipping them when you have FAQ content is leaving a low-effort signal off the table.
One more thing has started to matter, in ways most agencies are only beginning to measure: brand entity recognition. If your brand isn't a recognizable entity to AI systems (no Wikipedia presence, no Wikidata entry, inconsistent NAP across the web), you're harder to cite with confidence. Building entity recognition is slow. Annoyingly slow, sometimes. But the benefits compound across both GEO and traditional SEO.
What agencies specifically need to do differently
If you're running an SEO agency, the challenge isn't only technical. It's how you explain this to clients.
Clients who hired you to improve rankings need to understand that rankings are now one of several visibility metrics, not THE metric. You need to start tracking AI visibility separately. How often does the client's brand appear in ChatGPT or Perplexity responses for target queries? There are emerging tools for this. AI Rank Tracker, Search Response, and Profound are early-stage but getting more useful. Manual spot-checking is also just honest work.
The agency value proposition has to shift too. "We'll get you to page one" is a weaker promise when page one increasingly means an AI answer box that doesn't send clicks. The new pitch sounds more like "we'll make you the source AI systems cite, and the brand users recognize when they search for themselves."
That's harder to sell. I don't love that, but it's the accurate pitch.
Agencies that figure out how to audit for AI visibility, measure it consistently, and charge for that work will have a service that's genuinely different for the next few years. The ones that don't will keep selling something that's quietly losing power.
What I think is actually going to happen
My read: traditional SEO doesn't disappear. Local SEO, product SEO, and branded search stay relevant. But informational content (the "what is," "how to," "best of" content that's been the backbone of most content marketing programs for a decade), that's going through a real shift.
The volume of AI-answered queries will keep climbing. CTR to organic results for those queries will keep dropping. The content that does get surfaced in AI answers will increasingly belong to sources that figured out early how AI systems pick and cite content.
Here is the uncomfortable part: being "good at content" may not be enough. You can write a useful page, rank it, and still lose the AI answer to a page that is easier to quote, easier to verify, and clearer about who is making the claim.
There's also the flip side. Brands that get cited consistently by AI systems build a different kind of trust. When ChatGPT recommends your methodology or Perplexity cites your research, that's a signal that compounds. Harder to build than a backlink, but harder for someone else to replicate too.
It works.
FAQ
What's the difference between GEO and traditional SEO?
Traditional SEO optimizes content to rank in search engine results pages (SERPs) based on factors like backlinks, domain authority, and keyword relevance. GEO (Generative Engine Optimization) optimizes content to be cited, summarized, or recommended by AI-powered answer engines like ChatGPT, Perplexity, and Google AI Overviews. These are systems that synthesize answers instead of returning a list of links.
Does GEO replace SEO, or do they coexist?
They coexist, at least for now. GEO and SEO share foundational principles (quality content, E-E-A-T signals, structured data), but GEO adds requirements around quotable content structure, named sources, and entity clarity. Most agencies need to run both in parallel instead of treating them as alternatives.
Which AI tools matter most for GEO visibility right now?
In 2025–2026, the most consequential surfaces are Google AI Overviews (highest volume), Perplexity (high engagement among technical audiences), and ChatGPT browse mode. Claude's web search is growing. For most brands, Google AI Overviews should be the first priority simply because of search volume. Perplexity is disproportionately influential with developer, researcher, and SaaS buyer audiences.
How do I know if my content is being cited by AI systems?
Manual testing is the most accessible starting point. Run your target queries in Perplexity, ChatGPT, and Google AI Overviews and note which sources get cited. Tools like Profound and Search Response offer more systematic monitoring.
This is still an emerging area. There's no mature GA4 equivalent for AI visibility yet, which is part of why it's worth tracking now, before it gets crowded. My take: the teams that start logging this manually now will understand the market before the dashboards get polished.
We run GEO audits for clients at webcorelab.com.