How I Measure Traffic From AI Tools Like ChatGPT (And What You Should Know)
Hi, I'm Josh, Managing Director at Peaco Marketing.
Like a lot of people in digital marketing right now, I’ve been thinking a lot about AI. Not in the “robots are taking over” way, more in the “how are tools like ChatGPT actually affecting our clients’ traffic and leads?” way.
One question I’m hearing more often is:
“Can we actually track traffic coming from AI tools like ChatGPT?”
The short answer: kind of. It’s not as straightforward as tracking Google Ads or organic search, but there are a few smart ways to get a handle on it. In this post, I’ll walk you through how we’re doing it at Peaco and what you can steal for your own business.
First off, why is this tricky?
Unlike a web browser or a social media app, ChatGPT and other AI tools (like Bing Copilot or Google Gemini) don’t always show up clearly in your analytics.
If someone clicks a link in ChatGPT, there’s usually no referrer passed along, which means in Google Analytics it just looks like Direct traffic. It’s the same bucket as someone typing your URL in manually or using a bookmark. Pretty useless when you’re trying to attribute results.
And sometimes, people don’t even click. They just read a summary, get their answer, and move on, which makes the whole idea of “tracking AI traffic” feel like chasing shadows.
But over the last few months, I’ve found a few things that actually work.
1. UTM Tags Are Your Friend
Anytime we’re trying to include a link in AI-generated content, whether it’s for our own blog or a client’s FAQ, we use UTM parameters as well as simply checking GA4 sources/mediums when more fitting.
Here’s what that looks like:
https://yourwebsite.com/your-page?utm_source=chatgpt&utm_medium=ai&utm_campaign=chatgpt-referral
If someone clicks that link, we can track it clearly in Analytics. It’s a small change, but it gives us solid data we can work with.
We’ve even tested this by dropping UTM-tagged links into our own prompts inside ChatGPT. When those links get clicked, they show up right where we want them to, no guesswork.
2. Custom Landing Pages Help
We’ve also created a couple of specific landing pages for certain AI-use cases. For example, we’ve written answer-style content that we want AI to pick up on, things like:
“How much should SEO cost in Yorkshire?”
“What’s the best way to structure a cold email campaign?”
When we promote these via AI tools or forums, we monitor the page performance closely. If we see a spike in visits without any clear referral, it’s a good bet ChatGPT or Bing is involved.
We cross-check that against tools like Bing Webmaster Tools and Google Search Console for Good Measure, and tools like SEMrush can help track it all too.
3. Checking the Server Logs (Yes, Really)
This one’s a bit nerdy, but stick with me.
We started looking at our server logs to see if AI crawlers are sniffing around. You won’t see ChatGPT itself, it doesn’t crawl the web, but you might spot bots like:
GPTBot
CCBot (from Common Crawl)
Bingbot (Bing powers ChatGPT browsing, remember)
If we see these crawling a client’s site, we know it’s being picked up in some form by AI models, even if the traffic doesn’t come right away.
4. Just Ask People
I know this sounds simple, but we added a question to our lead forms that says:
“How did you hear about us?”
Recently, one came through with the answer:
“Asked ChatGPT for a local SEO agency.”
That was the first time I knew for sure someone had found us that way. Now we keep an eye out for it.
So if you’re generating leads online, especially if you’re in a niche market or B2B, don’t be afraid to just ask the question. People are surprisingly honest about it.
5. AI Mentions Still Matter (Even If There’s No Click)
Not every AI reference leads to traffic, but it still affects how people see your brand. That’s why we use tools like Brand24 to track mentions, even outside of social media.
We also run periodic prompts in ChatGPT like:
“Who are the best SEO agencies in Yorkshire?”
“Where can I get help with PPC in Hull?”
If we don’t show up, we dig into why not, and make sure our site, content, and structure are in good enough shape to be pulled in next time.
The Bottom Line
If you’re trying to track traffic from ChatGPT or AI tools, don’t expect perfect data. But do start building a system.
Use UTM links. Create helpful, structured content. Watch for strange spikes in Direct traffic. Ask your leads where they came from. And start thinking of AI as a real, trackable channel, not just some mysterious black box.
It’s still early days, but we’re already seeing signs that AI can send real traffic and real business. And as more people start using these tools instead of Google, it’s only going to grow.
If you want help figuring this stuff out or just want to work with a team that gets how fast digital marketing is changing, drop us a message. I’m always happy to chat.
