I spent $500 to check if SEO is dead

AI has killed SEO and content websites no longer make sense because now chatbots give you all the answers. That’s what I’ve been hearing over the last few months. And yet, despite all the doomposting, millions of people still search things like “how long to boil potatoes” on Google every single day instead of asking ChatGPT.

Since I consider myself quite a skeptical person, and I like to validate hypotheses with data and experiments, I decided to prove whether this is actually true or not.

In the past I’ve managed several niche content websites, ranking them in the top positions on Google and monetizing them through Adsense. I ended up selling some of these websites after some time. My goal now is to buy a few websites with modest but steady traffic and take them to the next level by increasing both visits and revenue to see whether this is still possible in 2026, right in the middle of the AI boom.

My current hypothesis is that AI did not kill SEO. It simply raised the quality threshold. Small websites are still alive, but low-quality websites are probably doomed.

The Websites

I acquired 3 websites in 3 different languages within the same niche: cooking.

I chose these websites for several reasons. First of all, since they are in different languages, they allow me to test different local SEO strategies and compare how Google behaves across different internet ecosystems. Small-language internet markets are especially interesting to me because they currently seem less saturated by AI-generated content.

Secondly, all the websites follow a similar structure, so in the end they don’t require much more work than if it were just a single website.

As for the numbers, they receive around 5,000 monthly visits combined. The main website concentrates 50% of the traffic while the other two split the remaining half. Regarding monetization, all the websites were quickly accepted by Adsense and, at the moment of purchase, they generate around $15 per month.

Another positive aspect is that the websites have been running for around 3 years, which is good for SEO. One thing I’ve learned over the years is that the internet has a surprisingly long memory. Old websites, even mediocre ones, often have an advantage over brand new projects.

Why buying instead of building?

One question some people may ask is why I decided to buy existing websites instead of building completely new ones from scratch.

The answer is simple: ranking a brand new website on Google usually takes a long time and requires a considerable amount of initial work before you can even properly test whether your ideas work or not. You need to create content, build authority, wait for indexing, and slowly gain Google’s trust. Since the goal of this experiment is to test whether niche content websites can still grow in the age of AI, starting from zero would introduce too many additional variables and significantly slow down the experiment.

That is why I decided to buy websites that already have stable, although declining, traffic and try to improve them instead. This allows me to focus directly on optimization, UX, content quality and monetization rather than spending months trying to get a completely new domain off the ground.

Costs

The websites are all hosted under the same hosting service which costs $25 per year. On top of that, domain costs are around $12 per year.

Therefore, for the websites to break even they would need to generate at least $61 per year (without counting the acquisition cost). With the current RPM of $1.93 we would need around 32,000 visits per year. In the future I may consider cheaper hosting alternatives to reduce costs even further, but for now my priority was stability and preserving the websites’ existing rankings as much as possible. Since the websites were already indexed and receiving steady traffic, I preferred keeping the infrastructure changes to a minimum after the migration to avoid introducing unnecessary SEO risks.

The goal of this experiment is not only to see whether it is still possible to increase traffic on a niche content website in 2026, but also to see whether this is still a profitable option for someone who wants to make some extra money.

A lot of discussions around AI and SEO focus only on traffic, but traffic alone is meaningless if the economics behind it no longer work.

The Plan

Right now the websites have a very basic appearance and very little content per page. The good thing is that each of them follows the same structure, and that there are around 200 URLs ranking on Google per website.

The first thing I did after the acquisition was making sure everything worked correctly after the migration, touching as little as possible and only fixing a few redirect issues. One thing I’ve learned is that small websites are often much more fragile than they seem. A few bad redirects or broken pages can destroy years of SEO surprisingly quickly.

After a few weeks of stabilization it is time to get to work.

As I said, the websites are almost identical, so I will focus on one of them first (the one currently receiving the least traffic) and then replicate the changes that work on the others. I’m doing it this way to avoid making mistakes and killing the SEO of the other two websites with more traffic.

Right now I have several ideas to increase traffic.

Increasing the length of the pages

First of all, focusing on the pages that already exist and expanding their content.

When I bought the website, each page had around 150 words. They were not written by AI, which is a good thing, but they were all quite similar.

The first step is therefore to increase the amount of content on each page.

I started with the pages receiving the most impressions, taking them from ~150 words to around ~1400. I used AI, other websites and YouTube to research the topics, but I wrote all the content myself so that it sounds as natural as possible and to make sure everything is coherent with the purpose of each page.

One thing that is becoming increasingly obvious to me is that modern SEO feels much more like systems engineering than blogging. Structure, clarity, page speed and information architecture matter far more than most people think.

This process takes some time, so I still have several pages left to update. Luckily I enjoy cooking and learning new things. In fact, this is my favorite part of building niche content websites.

Improving the design and UX

Another thing that needs improvement is the page design, which right now looks like a blog from the early 2000s. In fact, the whole website is basically HTML and CSS.

I added some JavaScript to create tools that can help the user. With Claude’s help I made the whole UI much more visually appealing. Now it looks more like a website from this decade.

I also optimized all image sizes, cutting loading times in half, which will help improve the website’s ranking on Google searches.

One thing that fascinates me about small content websites is how tiny technical improvements can sometimes have disproportionately large effects. The internet economy is much more fragile than most people imagine.

I also added an FAQ section at the end of each page and some quick-answer cards at the beginning, which search engines usually like.

Expanding the verticals

The website’s content is very focused on a specific cooking niche: cooking times for different ingredients. But I think it has potential to include other kinds of preparations as well as recipes or even kitchen tool recommendations. I have already included some of this content in the existing pages, while some of it will go into completely new pages, which will increase the breadth of the websites and their reach by creating new content verticals.

Increasing revenue

As I said before, one of the goals of the experiment is to see whether niche content websites are still profitable in the age of AI, and for that I need to increase the revenue of each website.

These kinds of websites usually have two monetization paths: ads and affiliate links. Right now I’m monetizing only through ads, although I plan to include Amazon affiliate links once I add the pages about kitchen tools. For now I prefer focusing on optimizing the pages that already exist. By increasing the amount of content on each page, I also increase the available space for ads. The website had manually placed ads, and I kept some of them while also enabling automatic placement for the rest. With Adsense automatic placement, AI is used to detect the best number, position and type of ad for each page. In my experience it usually works quite well, although you need to be careful with certain types of ads that can feel invasive for the user, such as some overlay ads. In any case, those kinds of ads can be blocked from the Adsense admin panel, but for now I’ve left everything enabled.

Conclusion

Now I need to keep applying these changes to the remaining pages. We’ll see whether these changes have any effect or whether AI has really killed SEO and people now search only through chatbots. Follow me to read future updates about this and other experiments.