In the past two months, I landed 5 new clients. Not from Google ads, not from Instagram, not from referrals. From ChatGPT.
Each of them said roughly the same thing: "I was looking for someone to build my website and ChatGPT recommended you." I didn't pay for any of it. I didn't run a campaign targeting ChatGPT users. ChatGPT simply read my website and when someone asked it for a web developer, it mentioned my name.
This isn't a fluke. It's happening right now across industries, and most businesses have no idea.
What actually changed
Not long ago, search was simple. You typed a query into Google, got 10 blue links, clicked on one. That model is breaking down.
According to a Semrush study from 2025 that analyzed over 10 million keywords, Google AI Overviews (the AI-generated summary answers that appear directly in search results) show up in around 16% of all desktop searches in the US. In July 2025 that number peaked at 25% before Google pulled back slightly. The direction is clear: Google is evolving from a search engine into an answer engine.
At the same time, ChatGPT has grown into something nobody predicted this quickly. According to OpenAI, ChatGPT now has 900 million weekly active users, confirmed in February 2026. That's more than 10% of the world's entire population asking questions to an AI every single week.
And they're not just asking about recipes. They're asking: "Who builds good websites in London?" "Which agency should I use for my marketing?" "Where do I find a reliable accountant in my city?"
Meanwhile, Google's global search market share dropped below 90% for the first time since 2015, according to StatCounter data from Q4 2024. It hovered around 89.5% throughout most of 2025. On its own, a sub-1% shift sounds trivial. In the context of a market Google had dominated completely for a decade, it's the first real crack in the wall.
How AI search actually works
When someone types "who builds websites in my city?" into ChatGPT, it doesn't just pull from what it learned during training. It uses its own crawler, OAI-SearchBot, alongside Bing's index. It reads pages, evaluates relevance, and synthesizes an answer.
There's one technical difference from Google that matters a lot here: AI crawlers don't execute JavaScript. Googlebot has gotten good at rendering dynamic content. OAI-SearchBot, GPTBot (OpenAI's training crawler), PerplexityBot and ClaudeBot do not run JavaScript at all. They only see what's in the raw HTML response. If your website loads its content via JavaScript, which is common in modern frameworks, AI simply cannot read it.
AI also has a strong preference for fresh, clearly structured content. It favors pages that directly answer specific questions over pages stuffed with keywords and vague value propositions.
A 2025 report from Previsible, which analyzed 19 GA4 properties, found that referral traffic from AI platforms (ChatGPT, Perplexity, Gemini, Claude, Copilot) grew by 527% year over year, jumping from 17,076 sessions in January through May 2024 to 107,100 sessions in the same period of 2025. These numbers are still small relative to Google's scale, but the growth rate is hard to ignore.
Gartner predicted in February 2024 that traditional search engine volume would drop 25% by 2026 due to AI chatbots. That prediction hasn't fully materialized yet, but the direction it pointed to is playing out in real traffic data right now.
What this means for your business
When I looked into why ChatGPT recommended me to those five clients, a few things stood out.
My site was readable by AI. It uses server-side rendering, so the content exists in clean HTML. No JavaScript-dependent content that disappears for crawlers.
I had specific, direct content. Not generic copy like "we build high-quality websites." Instead: what I do, who I do it for, rough pricing, and how the process works. ChatGPT can extract a useful answer from that. It can't do anything with vague marketing language.
I had a llms.txt file. This is a plain markdown file proposed by Jeremy Howard of Answer.AI in September 2024. You place it at the root of your site and it tells AI models: here's what this site is about, here are the most important pages. Think of it like robots.txt but for AI rather than search engine crawlers. It's not a guaranteed ticket into AI recommendations, but it signals that your site is set up for it.
I had structured data (schema markup). When your content is marked up properly, AI understands it better. Who you are, what you do, where you operate, how to contact you.
Most businesses don't have any of this in place. That's the opportunity window right now.
What you can do today
These are practical steps that don't take long but can have a real impact:
1. Check your robots.txt
Go to yourdomain.com/robots.txt. If you see Disallow: / for GPTBot, OAI-SearchBot, or similar AI crawlers, you're blocking AI from reading your site entirely. Many hosting providers configure this automatically. To appear in ChatGPT search results, OAI-SearchBot needs to be allowed.
2. Test whether your site works without JavaScript
Open Chrome DevTools (F12), go to the Network tab, check "Disable JavaScript," and reload the page. Can you still read your content? If the page is blank or broken, that's what AI crawlers see too. Solutions include server-side rendering, static HTML export, or prerendering.
3. Add a llms.txt file
A simple markdown file at yourdomain.com/llms.txt. Include your site name, a short description of what you do and who you serve, and links to your most important pages. You can find the format and examples at llmstxt.org.
4. Add or improve your structured data
Schema markup for LocalBusiness, Person, or Service helps AI understand exactly who you are and what you offer. If you don't have it, adding basic schema is a few hours of work with measurable payoff, not just for AI visibility but for Google as well.
5. Write content that answers real questions
Skip the "we're a dynamic team with innovative solutions" copy. Write like this instead: "We build WordPress websites for small businesses. Projects typically take four to six weeks. Pricing starts from X." AI looks for direct answers to specific questions. The more directly your content answers something, the more likely it is that AI will cite it.
If you want to understand how your site tracks conversions coming from AI channels specifically, take a look at my web analytics service. And if you need help optimizing for search, including AI search, the SEO optimization service covers that.
To wrap up
This is not the end of Google. Google still processes an enormous volume of searches and will continue to do so for years. But a parallel discovery channel is growing fast, and most businesses aren't set up for it yet.
The businesses that prepared early have an advantage. Not because they spent more money. Because they made a few technical and content decisions before it became obvious that they needed to.
For me, those decisions brought in 5 clients over two months without a single paid ad. I'm not claiming it'll work the same way for everyone. I'm saying it's worth taking seriously.
If you have questions or want to talk through your own setup, the contact form is right there.
Sources:
- ChatGPT 900M WAU: OpenAI, February 2026 / TechCrunch
- Google AI Overviews ~16%: Semrush AI Overviews Study, November 2025
- Google below 90% market share: StatCounter, Q4 2024 / Search Engine Land
- 527% AI traffic growth: Previsible 2025 AI Traffic Report / Search Engine Land, August 2025
- Gartner prediction -25%: Gartner press release, February 2024
- OAI-SearchBot: OpenAI Developers
- llms.txt: Jeremy Howard, Answer.AI, September 2024