I had one of those moments today where something I’ve been talking about for months suddenly stopped being theoretical.
I was looking through the analytics for one of the websites I work on, Sahara Sands, when I spotted a visitor I knew wasn’t me.
The traffic source was ChatGPT.
That alone was interesting. But then I opened the full visitor journey.
And that was when it became really interesting.
This wasn’t just a referral
At 12:33:46, the visitor arrived on a page about Plateau Breakers.
The referrer was:
https://chatgpt.com/
The landing URL even contained:
utm_source=chatgpt.com
So this wasn’t somebody finding the site through Google, TikTok, social media or a direct visit.
They had come from ChatGPT.
What happened next was the fascinating part.
- They landed on the Plateau Breakers page.
- They visited Best Sellers.
- They opened the Guides section.
- They moved into the Achieving Your Tan guide.
- They read an article called “Why Has My Tan Stopped Getting Darker?”
- They searched the website for “Plateau breaker”.
- They opened the Prismatica product page.
- They clicked through to shop.
The entire journey took less than two minutes.
That looks remarkably like somebody arriving with a question, researching the problem, discovering a possible solution and then moving into a commercial action.
This is exactly what I’ve been building for
I’ve been putting a lot of work into the structure of Sahara Sands recently.
Instead of thinking purely in terms of traditional SEO pages and keywords, I’ve been building around questions.
Things people genuinely ask.
Why has my tan stopped getting darker?
What does a tanning accelerator actually do?
What is the difference between an accelerator and a maximiser?
What should you use if your tan has reached a plateau?
Those answers then connect into larger guides, product categories and individual products.
Part of the thinking behind that is AEO, or Answer Engine Optimisation.
If somebody asks an AI assistant a question, I want the website to contain an answer that is useful enough, clear enough and structured well enough to become part of that discovery process.
Today I got to see what that can look like from the other side.
ChatGPT did the discovery. The website did the selling.
I think that is the most important part of this entire journey.
ChatGPT appears to have been the discovery layer.
It got the visitor onto the website.
But once they arrived, the website took over.
The visitor explored related information, moved through the content structure, used the site search, discovered a product and eventually clicked through to shop.
That distinction matters.
There is a lot of discussion at the moment about AI replacing search engines, websites losing traffic and users getting their answers without ever clicking anywhere.
Some of that will undoubtedly happen.
But this journey shows another possibility.
AI can also become a new source of highly relevant discovery traffic.
The website still matters because when somebody wants more detail, wants to compare options, wants to understand a subject properly or wants to buy something, there needs to be somewhere for them to go.
I’m not declaring Google dead
This was one visitor.
I’m certainly not going to pretend one ChatGPT referral proves that the entire internet has changed overnight.
It doesn’t.
But it does prove something much smaller and, to me, much more useful.
The mechanism is real.
People can ask an AI assistant a question, discover one of my websites through the answer, arrive on relevant informational content and then move from that content into product discovery.
I’ve now seen it happen in my own analytics.
That is very different from reading an article predicting that it might happen one day.
The analytics need to change too
This also changes the way I want to look at traffic.
Traditionally I’d be looking at Google, direct traffic, social traffic and referrals.
Now I want AI referrals treated as their own source.
I want to know which questions appear to generate AI traffic.
I want to know which pages those visitors land on.
I want to know what they read afterwards.
I want to know which products they discover.
And most importantly, I want to know how often they eventually click through to shop.
If enough journeys like this start appearing, that data can then influence what gets written next.
That creates a feedback loop.
Build useful answers. Measure what AI sends traffic to. See what those visitors actually want. Build better answers.
The future of search might look very different
For more than twenty years, building websites for search has largely meant thinking about Google.
Keywords. Rankings. Search results. Click-through rates.
Those things are not disappearing tomorrow.
But there is now another route into a website.
A person asks a question in natural language.
An AI helps them understand the subject.
The AI points them towards a useful resource.
The website continues the journey.
And in this case, that journey ended with a product and a shop click.
I happened to have built enough tracking to see every step of it.
At 12:33 this afternoon, I think I watched a tiny example of what search on the web is going to become.
And I suspect I’m going to be seeing a lot more of it.
