§ SEO · AI Search
SEO is dying. And the people selling “AI search strategy” are lying to themselves.
“AI is eating your clicks, but the traffic you keep converts 4-5x better.” The first part is true. The second is where the story falls apart, so I read the studies and ran the numbers.

SEO is dying. And most people selling "AI search strategy" are quietly lying to themselves about the one number holding their story together.
The story goes like this.
"Yeah, AI is eating your clicks. But the traffic you keep converts 4-5x better. Fewer visitors, richer ones. You're fine."
The first part is true. The second part is where it falls apart.
So I went and read the studies. And ran some numbers myself.
The most serious one is from Amsive. 54 sites, proper statistics. Organic converted at 4.6%, AI at 4.87%.
A peer-reviewed study by Kaiser and Schulze went further. 50,000 ChatGPT purchases against 164 million normal ones, across 973 ecommerce sites. AI visitors looked, showed interest, then converted below every channel except paid social.
So where does "AI converts way better" come from?
Soft stuff. Email signups, free trials, and self-reported data from companies that sell GEO tools. The number looks great when a conversion means someone handed over an email address. It falls apart when a conversion means someone actually spent money.
Here is an example.
Take a keyword with 100,000 searches a month, and say you rank number 1.
- With no AI, that is about 28,000 clicks. At a 2.8% conversion rate, that is 784 conversions.
- Today AI already eats into that. You are down to around 580.
- Now push to a world where AI runs 90% of search. The clicks crater. The answer sits on top and nobody scrolls. Even being generous and assuming AI converts better than 2.8%, you land around 150 conversions.
Same number 1 ranking the whole way. You did nothing wrong, and still lost roughly 80% of your business. That is what "just evolving" actually looks like.
Now the part that makes me nervous.
Right now GEO is basically SEO with extra steps. Every "GEO expert" selling a brand new playbook is mostly reselling the old one.
That is why everyone says SEO is not dying, it is just evolving. But the math above already breaks that. Cited or not, the conversions slide. Sooner or later clients figure out they are paying for rankings that no longer pay back.
And it gets worse. There is no law that says a model needs your sitemap.xml. Or your title tags. Or your meta descriptions. Or your backlinks. Those exist because crawlers needed them. The moment models get their information another way, through licensing deals, direct data feeds, or their own agents fetching it, every signal SEO is built on turns into decoration.
And here is the part that stings. SEO was the last channel where the little guy could win on merit. You could outrank a billion dollar brand by being better, no budget required. A world of licensing deals and 3 or 4 named brands per answer has no room for that. It is over for the little guy.
People have called the death of SEO for 20 years, wrong every time. I rolled my eyes at all of them.
This is the first time I think the naysayers might be right.
This time, I would not bet on it surviving.
§ Sources

§ About the author
Cristian Balau is a marketing & growth executive who builds with frontier AI. He helped scale Mondly to 150M+ users (acquired by Pearson) and is COO at Cratox AI.