State of Search in 2026: GEO, AEO and the Future of SEO

An honest look at GEO, AEO and AI search in 2026: what the traffic data shows, which tactics work, and why SEO is still the foundation.

Written by Nik

Founder & CEO, Get Stuff Digital

Published
13 min read

Since Google started weaving AI answers into search with AI Overviews in 2024 and rolled out AI Mode in 2025, a lot of people in our industry have lost sleep over one simple question: will LLMs replace traditional search engines?

Now that we have much more data and experience, it's time to revisit it.

For a lot of experts and professionals, 2025 was rough. Between days of worrying whether our jobs would still exist in 2026, overnight sellers of new services like GEO, and companies raising millions in VC funding to build tools that track probability, it's fair to say 2025 was an interesting year.

As you’d expect, it wasn’t easy for me either. Selling SEO was already difficult, and once AI search went mainstream, finding clients felt almost impossible. That being said, after many years of running an agency, I knew that difficult periods always bring new challenges, but also new opportunities. This was one of them.

More importantly, I knew the bubble around AI and LLMs would burst sooner or later and expose its weaknesses. Most of our clients are early-stage startups at the forefront of AI, and I felt we were at the beginning of something great.

Even before the craziness started, we were getting questions like "Why should we pay someone to write content when we have AI?" and "Why should we pay for SEO when we have AI?". And even when we didn’t get them upfront, we knew that’s how most of the people were thinking anyway.

On top of that, a lot of "AI bros" were sharing miracle AI agents that would supposedly steal your competitors' SEO tactics in a few clicks, hijack their traffic and make you millions overnight. That only built more animosity toward LLMs and their role in search.

I never dodged those questions. We knew what we were doing, and we kept doing it year after year, building results for each client step by step. Our game plan didn't change overnight in 2025. We waited until we had something solid to draw conclusions from.

By the end of 2025, we did:

Is SEO dead? Clearly not.

GEO or AEO? We still don’t know.

Is SEO at the foundation of both? Yes, it is.

Everyone Loves a Shiny New Thing

So why did GEO and AEO suddenly become the new wave? Cough HYPE cough. Everyone loves something new, everyone loves being a first mover, and everyone loves showing up on new channels.

But SEOs, there's no need to panic. Look at the giant companies that cut large parts of their workforce and replaced them with AI, simply because their stock price went up when they looked "super modern."

"Cut 5,000 people in favor of AI" sounds like an amazing move, but only when you begin with the premise that it’s beneficial for the company. Newsflash – it wasn’t and it didn’t take long for reality to hit everyone like a brick wall.

Remember Klarna and their revolutionary multi-tasking, ultra-capable, hyper-advanced, T-1000-levels-of-competency AI assistant that was carrying out 2.3 million conversations and doing the work of 700 human operators? The company announced it in February 2024. By May 2025, CEO Sebastian Siemiatkowski admitted that the whole automation stuff had “gone too far”. So, you know, they started hiring people again.

The moral of the story: a shiny wrapping doesn’t mean that the candy tastes good.

What the Data Actually Shows

GEO and AEO are very real. They opened new angles in how people search, find information and, logically, make decisions. But the impact is still at a very early stage, and that's the part many people struggle to understand.

All the numbers tell the same story: LLMs still send only a tiny share of total traffic. Across the wider web, AI assistants account for well under 1% of visits, somewhere between 0.1% and 0.43% depending on the study. Ahrefs measured 0.43% across more than 115,000 sites in September 2026.

Even for tech companies, which get more AI traffic than most, it’s typically around 2-3%.

So I need to be blunt and ask every CMO, VP of Marketing and decision-maker reading this: would you bet everything on a channel that sends a few percent of your traffic at best? Probably not.

In case you still don't believe it, here are a few of our results from 2025 onward:

Search Console for a client site: clicks and impressions climbing, 4.83K clicks and 485K impressions at a 7.2 average position
Search Console comparison for a client site: clicks and impressions well above the previous period
Search Console for a client site: 4.04K clicks and 771K impressions, with impressions rising sharply toward the end
Search Console comparison for a client site: clicks and impressions above the comparison period throughout
Search Console for a client site: clicks building steadily, 3.48K clicks and 105K impressions at a 3.3% CTR

GEO vs AEO vs SEO: What We Actually Do

Firstly, you can find much better technical content on this topic from Dan Petrovic, a legend in the SEO field, Mike King and others. Secondly, let’s talk about what we actually do in-house.

Simply put - nothing new really, compared to what we were doing pre AI storm. At Get Stuff Digital, we've always practiced SEO our own way:

  • Entity over keywords.
  • Coverage instead of blatant content.
  • Authority over marketplace links.

Coming from a guy who built his early career and spent years in the black hat industry, are there other ways to rank websites in 2026? For sure. But here, I'm only talking about the white hat methods we use for clients. I'll point you to some black hat people later on.

The Basics Still Win

On-page, tech, off-page, content. Just the basics. Semantics, internal linking, coverage – all these still work, as long as they sit inside a real content strategy.

However, while ranking well also helps you get cited by LLMs, this is no longer a guarantee.

In mid-2025, Ahrefs found that 76% of the top AI Overview citations came from pages ranking in Google's top 10. By March 2026, with a bigger dataset and Gemini 3 powering AI Overviews, only about 38% of citations came from the top 10, with another 31% from positions 11 to 100.

In other words, ranking still gets you in the room, but it won't land the citation on its own. That's where the off-site work comes in. (And since I always love a good Reddit thread, here's one tackling the same issue.)

Beyond the basics, we focus on a few things that also line up with Semrush's analysis of the most-cited domains in AI:

  • Being present on the platforms your ICP actually uses: forums, communities and social platforms.
  • Collecting user reviews.
  • Structuring FAQs so they're genuinely helpful for your ICP.
  • Good schema markup coverage.
  • Getting featured in third-party "best of" lists and earning brand mentions and authority on trusted sites.

And then there are the "new findings" that keep popping up in my feed:

  • llms.txt: The biggest distraction ever, at least for search. Google itself has said it doesn’t support llms.txt and isn’t planning to (Search Engine Land). We still add one during setup because it takes minutes, but if someone sells it to you as a GEO strategy, run.
  • Schema: Super good (WOW).
  • Listicles: Super good (Just make sure it’s someone else’s list).
  • Great content and authority matter (Oh, you must be joking…).
  • Ranking well gets you cited more (Does it, though?).

You get the point. I'm also aware that many SEO agencies have done poor work in the past, so for them, GEO might look very different from SEO. They'll end up selling those "juicy impressions" as the main KPI without even understanding what they mean.

Good GEO is Good SEO Done Properly

Here's the simplest way I can put it: good GEO isn't the same as what most standard agencies call SEO. But good GEO is good SEO, when SEO is done the way it should be. SEO is the foundation, so great GEO starts by building great SEO for your company and then builds on top of it.

Here are a few examples where we did only SEO and still saw a strong effect on citations:

LLM traffic dashboard: 105 sessions from AI assistants, 57 of them engaged
Daily sessions from AI assistants: none before October 2025, then regular visits of up to five a day

To be fair, we've always approached SEO a bit differently. I've always seen it as a multidimensional channel that feeds the rest of your marketing, not something that lives on its own. It should provide data for your other channels, be tied to clear KPIs, and work as a brand visibility and lead gen channel combined.

Maybe I just see things differently than the agency that'll send you the next invoice in 10 days for three blog posts and five links. Do we do those things too? Sure. But everything we charge for is tied directly to the KPIs we set and the results you achieve.

That's also why SEO struggled with its reputation for so long, with a lot of scammers in the mix. And a lot of the GEO "bros" landing in your inbox today fall into the same category.

Are AI Visibility Tools Worth It?

Oh, I’m glad you asked.

AI visibility tools are raising money like it’s 2021. The biggest one in the category is now valued at $1.8 billion (Value Add VC). Another closed its seed round just five weeks after its pre-seed (Surmado). All in, the ten best-funded GEO and AEO startups have raised about $604 million. You’ve got to love the AI bubble, am I right?

But what do those tools actually do? Well, you might not have seen this coming, but they essentially just track probability.

Why pay for prompt tracking when every answer is personalized based on the conversation a user previously had with the LLM? Sure, if you want to simulate a clean slate and see what a prompt returns without prior context, it can make sense. But does that really justify the cost? I'm not so sure.

That said, we do track prompts for our clients with our own in-house tool, because it’s useful for seeing how you and your competitors show up over time. We just treat it as a benchmark, not gospel.

Free ways to track your AI traffic

Before you pay for anything, set up the free options:

  • GA4 plus Looker Studio. In GA4, create a custom channel group for AI traffic, using a referrer rule that matches sources like chatgpt.com, perplexity.ai, gemini.google.com, copilot.microsoft.com and claude.ai. Plug it into a Looker Studio template and, voilà, you have a free tracker for AI referral traffic. It won’t capture AI Overviews, and it’s not perfect, since some AI visits show up as direct traffic, but it covers the chatbots.
  • Bing Webmaster Tools. Bing now has an AI Performance report that shows how often Copilot and Bing’s AI summaries cite your pages, which pages get cited and the grounding queries behind those citations (Practical Ecommerce). It shows citations, not clicks, and notice what else it shows: grounding queries, meaning the searches the AI runs behind the scenes, not the prompts people actually type. If Microsoft, with all its data, can’t show you those, be skeptical of any tool that claims it can. What those tools show you are prompts they ran themselves. Useful, but not the same thing.

I also tested a bunch of those paid tools myself (I won’t name any), and their “effect” on your position in LLMs is ridiculous. If you've followed my point so far, I won’t repeat myself, but I’ll put it like this:

Basic GEO advice from a tool built by grads with no SEO experience won't magically get your company cited more.

Naturally, this didn’t stop people from trying and now we’re seeing the effects: Google nuking websites built specifically for GEO purposes.

First, let’s take a moment to admire the slippery slope (pun intended) on full display:

Ahrefs estimate of organic traffic for a site built for GEO, falling from about 16K to under 2K between August 2025 and February 2026

Now let’s take a few more moments to breathe in the irony of optimizing the website for more LLM citations just to obtain the exact opposite result:

AI citations by platform for the same site: AI Overview down 54, ChatGPT down 44, Gemini down 4 and Perplexity up 42

I won’t lie, this is an impressive level of failure. I mean, who would have expected that doing bad SEO and resting your entire growth strategy on GEO (as if GEO and SEO are 2 entirely different ecosystems) would drive your website into the ground?

But, cynicism aside, the point here isn’t that you shouldn’t use AI in 2026. On the contrary, the opposite is true: AI has become almost indispensable in the current SEO environment. The difference is in how you use it.

Here’s how I see things from my side:

  • Feed LLMs with data from your SEO tools, let them make summaries, let them pull cross-analysis, play with numbers - then take this and make conclusions by yourself.
  • Use AI to enhance the content that you research and write yourself.
  • Use AI for grammar polishing, cross-analyzing your content with other online sources, and analyze scripts, forums, and testimonials.

You can probably see the pattern here. AI isn’t the engine behind your growth strategy, you are. AI is just the wrench you use to tweak your approach and methods.

Regarding content for GEO - say more with less. I always loved and practiced that. A proper content rework is often a faster win than publishing something new.

But, if you go with semantics and entity coverage, it all comes down to (again) good SEO practices: don't target and spam keywords blatantly, don't fill content with keywords just because you need to target them (we are not in 2010 anymore…)

Are Black Hat Methods Coming Back?

I think black hat methods are definitely making a comeback, especially in the GEO and AEO space.

If you want to learn more about these methods, check out Charles Floate, the GOAT in the ecosystem, covering everything from parasite SEO to pSEO and expired domain strategies, to name a few.

For GEO, it's crazy easy: just spam listicles everywhere, because LLMs are still guided by the quantity of mentions, not their quality. We even tested it ourselves early on, and it worked, which tells you everything about where AI search still is. I don’t like it, but, hey, it’s where we’re at.

To be clear, I’m not talking about earned placements. Getting featured in independent, editorial “best of” lists is legit, and it’s part of what we do for clients. The black hat version is publishing, or buying, dozens of lists that exist only to tell AI you’re #1.

The good news is that Google is finally catching up. Since May 15, 2026, its spam policies explicitly cover attempts to manipulate AI Overviews and AI Mode (PPC Land). And self-ranking listicles are already losing ground: Lily Ray found that AI Overviews cited 38% fewer of them in September than in June, and left the brand that published them out of the recommendation 83% of the time (Lily Ray).

So if you go down the black hat route, know that the window is closing fast.

Still, I think black hat methods for GEO will be common for years to come, because the results are so easy to get. That's why this isn't a typical article about how LLMs pull citations. It's a raw look at what actually drives results there, and it's still not pretty.

My Final Thoughts on AI in 2026

I really think AI is amazing for a lot of things. For search specifically, it opened new channels, which I find exciting. Vibe coding tools and AI agents that automate boring, repetitive tasks are also undeniably awesome and are sitting behind most of the SEO trends we’ll be dealing with for years.

But we also need to realize that AI is still in its early stages. Hallucinations are still a thing, and it's still far from real intelligence (how far, I can't say). Keeping that in mind, we need to learn how to use these tools properly.

If you're reading this, please don't fall for the snake-oil GEO "bros" and their miracle three-click setups on your favorite platform. That's not how SEO, AEO or GEO works.

What you should be doing instead:

  • Build on a solid SEO foundation. It's what both Google and LLMs rely on.
  • Dust off what's been sitting in the drawer. Schema, cough cough.
  • Treat marketing as omnichannel, with SEO as one part of it. Do everything for your ICP so your product can speak for itself. There can be thousands of AEOs, GEOs and SEOs, but word of mouth and product-led growth will never be beaten.
  • Tie every SEO effort to a clear KPI. Leads, newsletter signups, brand visibility, citations, traffic, even those horrible impressions (that's a joke). Without one, you're just falling down the rabbit hole.
  • Demand full transparency from your agency or consultants. At any moment, you should know where your money is going, what they're doing on your website and why. If you don't, start looking for another agency immediately.

That's my two cents, maybe a bit more given how long this article ended up being. Hope you enjoyed it.

FAQ

Is SEO dead in 2026?

No. LLMs still drive a small share of total traffic, a few percent at most, and under 1% across the wider web, while Google still sends most visitors. AI answers also rely on the same signals SEO builds, so strong SEO remains the foundation of AI visibility.

What's the difference between GEO and AEO?

In practice, very little. GEO (generative engine optimization) and AEO (answer engine optimization) both describe getting your brand cited or recommended in AI-generated answers. The industry hasn't settled on one name yet, but the work behind both is the same.

Should I pay for an AI visibility tracking tool?

Not as a standalone purchase. Start with the free options: a GA4 channel group for AI referral traffic and Bing’s AI Performance report for Copilot citations. Prompt tracking is useful as a benchmark for how you and your competitors show up over time, which is why we run our own in-house tracking for clients. But real answers are personalized, so treat the numbers as rough signals.

Not for search. Google has said it doesn’t support llms.txt and has no plans to, and AI crawlers rarely even request the file. It’s quick to add, so it doesn’t hurt, but your time is better spent on crawlability, schema and content quality.

Do self-promotional listicles still work for AI visibility?

Less and less. Google's spam policies now cover manipulating AI answers, and recent data shows AI Overviews citing far fewer self-ranking "best of" lists, often leaving the publishing brand out of the recommendation entirely. Earned placements in independent, editorial lists are a different story, and still carry real weight.