AI Has Changed the Rules: Which Business Models Are Dying and Which Ones Will Survive

Lately I've been observing my own search behavior, and what I see concerns me. My use of Google has dropped dramatically in recent months. If before I used it constantly for any technical or professional question, now I limit it to very specific cases and almost always end up on YouTube watching video tutorials.

For more complex text searches, I tell AI to search for me. To search through dozens or hundreds of pages, give me a summary of what it found, and then I go straight to the articles or forums I need. If it's verified information, I trust it to some extent precisely because it's mentioning the original sources.

With this behavior I'm visiting far fewer websites than before. When I need data I practically don't browse through dozens of pages, I go straight to the point. And if I've changed my habits, imagine millions of users. That has very serious implications for the traditional SEO model.

Two Types of Websites, Two Different Fates

I think it's important to distinguish between two categories because AI is affecting them in very different ways.

On one hand are transactional websites. Real estate portals, marketplaces, directories, any site with live content where people need to see images, review prices, use a search engine, sort results, or interact with a list of products. In this case, for now AI hasn't replaced anything and this type of project is still needed.

These websites have several revenue models. They can make money through advertising, whether Google AdSense or sponsorships, but above all they monetize when users pay to promote ads, to publish content, or when they buy the products advertised on the website.

That's why in these niches the business is going to continue working. While it's true that the advertising revenue branch could be diminished if people use AI to browse the web less, the business model remains solid. AI isn't going to kill this type of project in the medium term.

On the other hand are informational websites that are based on positioning through SEO, having lots of traffic, and monetizing through advertising, banners, or affiliate marketing. That's where the model breaks. If the user no longer enters your site because AI gives them the direct answer, there's no business.

The future of web development is going to polarize. On one side, more specialization in complex transactional websites where AI can't yet compete, with advanced filtering and external API integrations. Those projects will continue to exist and become more sophisticated.

On the other side, a shift toward developing internal infrastructures, control panels, intranets, and APIs that expose data so it can be consumed by AI models. The web developer of the future probably won't be building public websites as much as building the infrastructure that other AIs will use.

Affiliate Marketing: The Final Nail in the Coffin

I speak about affiliate marketing with firsthand knowledge because I've watched it decline in recent years. Amazon has been paying less and less commission on affiliate purchases, which was already a clear sign that this sector was in decline. There used to be websites cropping up like mushrooms everywhere, low quality, automatic, that didn't provide authentic product reviews but positioned themselves on Google and made money through commissions.

I know people in the sector who sold their websites years ago because they could see that the model would go down. And now with AI it's been the final nail in the coffin.

Think about the typical affiliate articles: which are the 10 best computers for X, the best coffee makers, the cheapest ones. With AI people no longer enter those blogs. They ask the model directly to do the comparison and tell them which products are best for their case.

It's true that AI will probably lie in some cases because it's based on those same blogs that contained invented information. But that no longer matters because the flow has been broken. AI gives you directly the product name and surely a link to the official website or Amazon, without affiliate code in between. The blogs lived on clicks inside the article, and that no longer happens. That's why in my opinion this sector is completely dead.

Tailwind and Other Obsolete Models

The Tailwind example is perfect for understanding another type of affected model. Adam Wathan, its founder, recently announced that he had laid off 75% of his employees. Pure irony because Tailwind is more popular than ever, but traffic to its documentation has fallen 40%.

Before, when a developer wanted to create a responsive navbar, they had two options: invest 3-4 hours programming it or buy the pre-made component in Tailwind UI. Now, with a 30-second prompt, AI generates that component for you. The framework is still being used massively, but the company that created it doesn't see a cent.

Tailwind's business model was viable and profitable before AI. Selling pre-made UI components made sense because it saved you hours of development. But with current models that product no longer adds value to the customer. The customer prefers to pay an AI subscription that allows them to generate anything rather than paying to have a package of pre-generated HTML as templates.

It's not an isolated case. Grammarly competes against ChatGPT's native writing capabilities. Productivity plugins for VS Code are being absorbed by Copilot. Stack Overflow has seen its traffic drop dramatically because developers ask LLMs directly.

The Core Problem: Broken Incentives

This destruction of business models has a common cause. The key for an LLM to be useful is the corpus it's trained on. And that corpus, until now, has been basically the entire internet: tutorials, technical documentation, Stack Overflow, blogs, open source repositories.

Why did all that content exist? Because of clear incentives. You wrote a technical tutorial and Google sent you thousands of visits. You answered on Stack Overflow and built professional reputation. You launched something open source and gained users, contributors, and eventually forms of monetization.

LLMs break that chain. They respond directly without sending traffic. The user never sees your name or your website.

Before, big tech companies won when you won. Google sent you traffic, Apple gave you customers in the App Store, AWS provided you with cheap infrastructure. There was a moment when the ecosystem grew alongside the platforms. Now the cycle is much faster: AI providers open up with cheap APIs, a third-party ecosystem is created, they learn from them, they integrate what's profitable in the next release, raise prices, lower quality, and everyone gets screwed.

Microsoft took decades to commoditize entire software categories. AWS takes years to clone successful products from its marketplace. AI providers do it in weeks.

If creating good content doesn't pay, the system fills with mediocrity. The problem isn't that AI writes worse, it's that it's destroying the ecosystem that produced quality content. The models are trained on knowledge accumulated over decades, but they kill the incentives that generated it.

Traditional SEO Is Dying

If search behavior has changed, how we position ourselves also has to change. Traditional SEO as we knew it is dying.

People no longer search on Google to get a list of websites and navigate through them. Now they ask AI directly. That means positioning is no longer so much in search engines, but in AI chats.

And that changes everything. Many people still say you have to keep publishing articles on your blog to position yourself. And it's a yes but no. What I'm seeing is that for AIs it's more important that your company brand is mentioned from other websites that give authority to your brand, than that you have 100 very well-positioned articles in SEO.

Your own content helps, of course, but it's no longer the main thing. The main thing is that you have a strong and known brand, mentioned by other authoritative websites. In this new landscape, what other websites say about you is more important than what you say about yourself.

And there's a key detail: that they mention you well. It's useless for other websites to talk about you if they're critical or negative mentions. AI interprets this and it influences the response it gives the user. If your brand is associated with positive experiences, you'll come out better in the responses.

My Experience: Drupal vs WordPress

My concern in this sense is relative because my work sector is different. In Drupal I'm more focused on large and complex projects, not on simple blogs that position themselves through SEO.

Most websites I work with are government portals, corporate portals, user intranets, pages where people register to vote on ratings, subscribe to events, make event payments and reservations. In the end user data is being managed, user payments are also being managed in several of the options I'm working with.

Yes, it's true that some of my clients have a large part of their income from advertising or affiliate marketing. But also a large part of clients don't depend on advertising or affiliate marketing, they simply have income from other areas that AI can't substitute for now.

That's why I see a clear difference of futures between the two ecosystems. WordPress websites that are more focused on blogs and making cheap cookie-cutter websites to try to position articles for SEO are a type of website that will be very affected and that have no place in the distant future.

In contrast, in my sector focused on Drupal we work differently for larger, more corporate websites, with many more functionalities than just making blogs for better SEO. Those projects will continue to exist and be necessary.

Impact on Juniors vs Experts

But this affects not only the type of websites, but also professional profiles.

Frontend

Today it's quite easy to generate templates with a more than acceptable visual style, not perfect, but more than acceptable with AI by anyone who knows how to use it basically.

Hyper expert frontend developers will continue to have work on more complex designs, but for simple designs frontend juniors have it quite hard to find work. A person with barely any technical knowledge will be almost the same thing.

Where there is opportunity is in design. A person with design knowledge will find it much easier to find work. AI doesn't have the same aesthetic criteria as a human designer, and there's still value there.

Backend

The same thing happens in backend with simple code generation. Pretty junior people have it complicated to enter the sector or to have a decent salary. AI already generates basic code that before required a junior.

In contrast, people with a lot of experience find it much easier to stand out if they use AI correctly. Experience is still necessary for complex architectures, non-trivial problems, difficult integrations. There AI is a tool, not a replacement.

The Conclusion

Those who don't know how to adapt and only know how to create simple websites, whether WordPress, Drupal or whatever, have their days numbered. But this doesn't mean it's the end, rather the start of a necessary specialization.

If you're junior, your way out isn't to compete with AI generating simple code. Your way out is to learn to use AI as a tool and specialize in what AI can't do: complex architecture, high-level design, non-trivial problems.

If you're an expert, your advantage is amplified. AI eliminates tedious work and allows you to focus on what really adds value.

The profession isn't dying. It's being filtered. And like any filter, it eliminates those who don't adapt and rewards those who do.

What to Do Today If Your Business Depends on SEO

If up to here this has all seemed very pessimistic to you, I understand. But it's useless to only see the problem if there are no solutions. Here are some practical recommendations depending on your situation.

If you have an informational website that lives on SEO traffic, you have two options: transform or die. Transforming means stopping thinking of Google as your main source of traffic and starting to think about how to make your brand appear in AI responses. That implies getting other authoritative websites to mention you positively, collaborating with media, participating in podcasts, getting quality backlinks. And it also means your own brand has to provide value that can't be replicated: own products, services, community, something beyond publishing articles.

If you have a transactional business, you're in a better position. But even so you should diversify your traffic sources. Don't depend only on Google. Invest in building an email list, create a community on social networks, develop own products. And above all, make sure your brand is recognized and mentioned positively on the web.

If you're a developer, start specializing. Simple WordPress for blogs is condemned. Learn about AI integrations, develop APIs, build complex data management systems. The future isn't in creating more public websites, it's in creating the infrastructure that will feed the AIs.

What worries me most about all this isn't so much the profession, but that we're consuming a finite resource, quality knowledge, without regenerating it. Big AI companies can sign multi-million dollar deals with publishers, but that only adds another massive cost to companies that already lose money and doesn't solve the fundamental problem.

The reality is that the SEO model based on informational traffic will probably never return to what it was. And that profoundly changes what type of projects make sense to develop in the medium term. Developers who understand this transformation and know how to adapt to new opportunities will continue to have work. Those who don't, will probably have problems.

We're in a moment of change, not end. But it's a change that many still haven't started to digest. And while they don't digest it, they'll keep building houses of cards on a beach where the tide no longer recedes.