What works now in SEO and AI (GEO) in 2026
By Christian Hernández · Published: · Updated:
Quick answer: SEO isn’t dead, but content as a product is in trouble: if your business lives on informational answers, AI already gives them away. What works in 2026 is focusing on transactional pages (products, categories, services), checking what Google shows for each keyword and adapting the format (video, images, Maps), building web “products” that AI can’t replace (tools, real calculators, templates), and earning mentions of your brand, with a clear value proposition, in the sources AI reads: Reddit, rankings, directories and media. GEO is built on SEO, and the same tactics that put you in AI answers can be used against you, so protect your brand’s results too.
Is SEO dead?
This article is based on an interview on the Campamento Web podcast, recorded in Andorra in April 2026. Emilio García talks with Jesús Arias, an SEO best known for running around 1,000 AdSense niche sites, about where SEO and AI search (GEO) really stand.
The mood in the industry isn’t great. They mention long-running media sites and well-known SEO events in Spain that have closed, and veteran professionals who are giving up. Jesús sees it differently: times of change bring opportunities, organic search isn’t going away, and his own sites still get a lot of traffic. The reason is simple, and it’s the thread running through the whole conversation: his sites don’t target informational keywords.
If your product is the content, you have a problem
If your business depends only on informational searches, it’s going to change, because Google’s AI Overviews, ChatGPT or any other AI already give that answer away. “What color is the sky?” no longer needs a website. That content has become a commodity.
There are exceptions where content still has value:
- Opinion and authorship: people want to read you specifically because of how you explain things or what you know.
- News and current events: to answer “what happened today”, AI still has to look it up on websites.
- Being the primary source: saying something new, with your own data or a different point of view, instead of rehashing what already exists.
And if your product isn’t the content (you sell shoes, software or a service), Jesús doesn’t see a problem. Part of the decision now happens in ChatGPT or in the AI Overview, so you may get fewer visits, but the people who arrive come ready to buy. Fewer visits, same sales.
Should a business still have a blog?
Emilio’s example: a shoe store or a psychologist. Jesús’s answer: having a blog doesn’t hurt, but it shouldn’t be near the top of your to-do list, because informational keywords won’t bring much traffic. First, work on everything transactional: product pages, categories and good copy, so that when someone searches “best running shoes for running in Madrid”, your page is the answer.
Look at the results page before you write
The most practical idea of the episode: Google already tells you, keyword by keyword, what format users want. Before investing in an article, search the keyword and look at what appears:
- Videos at the top? Make a video, not a text review. If all the results for a car’s reviews are videos, paying a writer for a written review won’t fit that search.
- A Google Images block? A blog post can be worth it even if it doesn’t rank, because the image inside it can.
- A Maps block? Work on your Google Business Profile.
- No AI Overview and classic links? There are still clicks to win with an article.
In other words, adapt to the market. If people prefer a 30-second video to a 500-word article, that’s what you should make.
Services and online stores: where to put the effort
For a business that sells services (an SEO consultant, a web designer), both agree the priority is service pages: one for each service you offer, expanded to as many long-tail variations as make sense, and even one per city or neighborhood if people search that way. It can be done with programmatic SEO.
They differ on how different those pages should be. Emilio recommends at least paraphrasing each one so it isn’t a copy-paste. Jesús says Google only treats content as duplicate when it’s exactly the same, so he wouldn’t worry if the pages are very similar.
My take: be careful with this one. Google’s spam policies call it doorway abuse when sites have “pages targeted at specific regions or cities that funnel users to one page”. A city page is fine when it’s useful for that city: real address or service area, local projects, prices, team, reviews. If the only thing that changes is the city name, it’s risky.
Jesús also sees the website as the center of operations: you may not get all your traffic through it, but people will go there to compare and to pay. Around it, create content for YouTube, TikTok or Instagram, depending on where your customers are.
Online stores have it easier, because so many of their keywords are transactional. When someone searches “buy iPhone cases”, there’s no AI answer to steal the click.
Building is easy now. Being found isn’t
One of the best reflections of the episode: creating a product, especially a digital one, keeps getting easier. With tools like v0 from Vercel or Lovable you can build a website or a SaaS in a couple of afternoons. They built their own course website with v0.
What’s harder than ever is distribution: getting your product in front of the right people. That’s why Jesús sees good SEO as a defensive moat. Anyone can copy your product; it’s much harder to copy your position in search.
Build your site with AI or with a CMS?
| Type of site | Their recommendation |
|---|---|
| Small site or landing pages | One-page or small corporate sites: have AI generate custom code. It’s faster, lighter and you can add exactly the features you want. |
| Calculators and tools | Custom code too, if they have few URLs. Inserting that kind of functionality into WordPress is harder. |
| Large or content sites | Dozens or hundreds of URLs and ongoing content: a CMS like WordPress is still very useful. Building the whole site with AI makes no sense there. |
Even with a CMS, Jesús creates the content with AI, and says it’s faster and much better than what he made four or five years ago. His workflow: he generates the content in bulk in Google Sheets, with an extension that connects to the OpenAI API (or any other model), and imports it into WordPress with a page generator plugin that also schedules posts and adds internal links.
Generating content at scale isn’t a problem in itself, but Google’s spam policies include “using generative AI tools or other similar tools to generate many pages without adding value for users” as scaled content abuse. Jesús’s point is that his pages are useful tools that solve the search intent; that’s the part that makes it work.
Are AdSense niche sites still worth it?
Jesús has been in the industry since 2018 and, around 2020–2021, found a type of site that hasn’t needed to adapt: transactional sites that AI doesn’t take traffic from. Would he recommend starting now? Only with the right knowledge, because niche sites aren’t made the way they used to be. Before, bad sites that didn’t solve anything could rank. Now the first question is: is your site a good product that gives users what they’re looking for?
If it is, advertising is one way to monetize it, not the only one. A product of your own may earn more, but it also brings more work (support, development). AdSense’s advantage is that it scales: you can monetize hundreds of sites just by adding a script.
“Call to click” sites
The ideal niche is one where the search intent can only be fulfilled on the website: downloading a template, generating a QR code, using a powerful calculator. The sweet spot is a product you can build with AI, but that search engines’ AI can’t solve by itself.
The risk: Gemini keeps adding features to Google, including more advanced calculators. A calculator where you enter four numbers and get four back is exposed. Jesús’s answer: pivot to something more, a small SaaS where users save their data, log in, keep a history or use an app. Again, the product can’t be just the content.
How they got their brands recommended by AI
This is the GEO part, and the most surprising one: they got brands created from scratch recommended by AI as the best option in their category in one or two weeks, and in some cases in 48 hours. Their method:
- Find out where the AI gets its information. For “who is the best…” questions, ChatGPT and Google search the web. Look at which pages they cite: often Reddit, Facebook, high-authority media, rankings and directories.
- Get mentioned in those sources. Even without a link: a mention of your brand is enough.
- Give that mention context. Not just your name, but why you’re the best: “the most complete SEO course”, “the most accessible AI newsletter”. They saw AI repeat those exact value propositions in its answers.
- Be most of what exists on the subject. If 100% of the content about a query says you’re the best, AI will say so. If it’s 5%, it won’t.
Competition matters a lot. Being “the best SEO in Spain” is hard because many people are already competing for it. Being the best SEO in a small town where nobody has written anything is easy. They also noticed that sites in the top positions of those rankings are mentioned more by AI, and that authority still weighs: a Reddit thread in third place can count more than a weaker page in first.
Do people even search for “the best…”?
Jesús adds a reality check: in Spanish, almost nobody searches “el mejor curso de…”; in English, “best…” searches are much more common, and many of these tactics were copied from the US. Where it does matter is in query fan-out: when AI breaks a question into several searches behind the scenes, it often adds “best” and the year. You type “running shoes”; the AI searches “best running shoes 2026”.
Emilio adds one more benefit, impossible to measure: when your logo or favicon keeps appearing as a source in AI answers, you’re also building your brand, even without a visit.
Owned, earned and paid media
To organize those mentions, Emilio uses the classic POEM model: owned media (your website, your blog), earned media (an interview, a mention in a ranking or directory) and paid media (a sponsored article). They’ve used all three. Owned media usually weighs the least, except when there’s nothing else written about you, because then it’s 100% of the content.
If you pay for a placement, Google asks you to mark those links with rel="sponsored" (qualify outbound links). And its spam policies cover third-party content published on a host site mainly to exploit that site’s ranking signals. A sponsored article in a real outlet, clearly labeled, is fine; content farms on other people’s domains are not.
The other side: anyone can do it to you
They even saw consultants who don’t exist being recommended by AI, because someone created enough mentions. The same goes in the other direction: if there’s nothing about your business and someone publishes five negative things, that’s what AI will summarize. And it does it faster than a person, because it reads every link and writes one paragraph at the top.
Jesús’s defense: try to control the whole first page of your brand search with your website, social profiles, review sites, interviews and articles in media. If you haven’t looked at what appears when someone searches “your brand + reviews”, start there. Always with ethics and transparency: getting real people to talk about you is one thing; inventing them is another.
What I take away from the interview
- Make your product something other than content: tools, services, products, opinion or original data.
- Prioritize transactional pages (products, categories, services) over the blog.
- Search every keyword before deciding the format: text, video, images or Maps.
- Use AI to build better and faster, but make every page useful.
- For GEO, find the sources AI cites and get mentioned there with a clear value proposition.
- Protect your brand search before someone else fills it.
And above all, the idea that came up again and again: the best GEOs will be the best SEOs. If you want to know how exposed you are, follow the 4-step GEO audit, and for the basics, how to increase your visibility in ChatGPT, Gemini and other AI.
The SEO base still comes first: titles, meta descriptions, H1s, image alt text and no broken links. basicseo.net checks all of that for free, page by page. Request free access.
FAQ
Is SEO dead in 2026?
No. Informational content is losing traffic to AI answers, but transactional searches (products, services, tools) still bring visits and sales, and AI tools read the same web that SEO works on. In the interview’s words, the best GEOs will be the best SEOs.
Should my business start a blog?
It doesn’t hurt, but it shouldn’t be the priority. Work first on product, category and service pages, and check each keyword in Google to see whether the results favor articles, videos, images or Maps.
How do I get AI to recommend my brand?
Find which pages AI cites for questions in your niche (often Reddit, rankings, directories and media) and earn mentions there, with context that explains why you’re a good option. It’s easier in niches where little has been written and harder where many brands already compete.
Can I create a landing page for every city?
Only if each page is useful for that city. Google’s spam policies treat city pages that only change the name and funnel users to one page as doorway abuse.
Sources
- Qué Funciona Ahora en SEO e IA (GEO) en 2026 - Entrevista a Jesús Arias (in Spanish) (Campamento Web, YouTube)
- Spam policies for Google web search (Google Search Central)
- Qualify outbound links for SEO (Google Search Central)
Leer en español: Qué funciona ahora en SEO e IA (GEO) en 2026