Is AI Sending You Customers That Your Analytics Credits to Google?

Quick Takeaways

1. 📈 People who had a brand recommended to them by ChatGPT were 2.5 times more likely to visit that brand's website within seven days, according to Similarweb research covering finance, travel and beauty.
2. 🔍 55.9% of those AI-influenced visits arrived through search, compared with 40.4% of visits with no AI involvement. In your reports they look like ordinary Google traffic.
3. ⏱️ AI-influenced visitors viewed about twice as many pages and stayed about twice as long as other visitors.
4. 🧾 Last-click reporting gives the whole sale to the final step, so the AI recommendation that started the journey gets no credit at all.
5. 🎯 The research shows an association, not proof that AI caused every visit. That is still enough to stop judging AI by the tiny "chatgpt.com" line in GA4.
6. ✅ Track four things together instead: branded search, your visibility in AI answers, direct traffic with engagement, and sales.

The customer you never see coming

Picture a woman in Townsville looking for a new robotic pool cleaner. She asks ChatGPT which ones are worth buying in Australia and where to get one. It gives her a short list and mentions your store. She does not click anything, because she is on the couch and the kids need putting to bed.

Two days later she types your business name into Google, lands on your site and buys.

Your analytics records a branded Google search and a sale. Google gets the credit. ChatGPT, which put your name in her head in the first place, does not appear anywhere in the report.

I see this pattern constantly when I talk to business owners about AI search. They open GA4, find a referral line from chatgpt.com showing a handful of visits a month, and decide AI is not worth worrying about yet. That conclusion is understandable, and the numbers below suggest it is often wrong.

What the research actually found

Neil Patel's agency, NP Digital, recently published a write-up by their Director of SEO, Rob Tindula, built on Similarweb research into what happens after an AI recommendation. The design is worth understanding, because it is more careful than most AI statistics doing the rounds.

Similarweb looked at people who asked an AI an industry question and got a specific brand recommended in the answer. They then watched what those people did over the following seven days. Importantly, they only included people who had not visited that brand before and had not named it in their own question, so the AI recommendation was the first real contact.

Three findings matter for any business owner.

People act on AI recommendations. Across finance, travel and beauty, people recommended a brand were 2.5 times more likely to visit that brand's website within a week than people who were recommended a competitor.

Most of them come in through search. 55.9% of the AI-influenced visits arrived through search engines, against 40.4% for visits with no AI involvement. That is the hiding place. A recommendation in ChatGPT very often turns into a Google search for your name a day or two later.

They are warmer visitors. AI-influenced visitors viewed nearly twice as many pages and spent roughly twice as long on the site as other visitors. They arrive having already been told you are worth a look, so they behave like someone shortlisting rather than someone browsing.

The article is honest that this shows an association. It does not prove that every one of those visits happened because of the AI answer. I think that honesty is worth keeping, and I will come back to what it means for how you measure.

Why your reports cannot see it

Most small business reporting works on last click. Whatever happened immediately before the visit or sale gets the credit. That model is simple and useful, and it was built for a world where the steps before a purchase mostly happened on websites you could track.

AI breaks that. The recommendation happens inside ChatGPT, Gemini or Perplexity, often with no click at all. When the customer finally does visit, they arrive through Google or by typing your address straight in. Your analytics records that last step faithfully and has no way of knowing what came before it.

There is a second effect that is easy to miss. Branded search, meaning people searching for your business by name, is usually read as a sign that people already know you. Some of that growth may now be the visible end of a journey that started in an AI answer. If your branded searches are climbing and you cannot point to anything you changed, it is worth asking where that awareness came from.

What to track instead

You do not need an expensive attribution platform to get a better picture. You need to look at a few signals side by side, every week, and watch whether they move together.

🔎 Branded search. In Google Search Console, filter your queries to ones containing your business name and watch the trend. Rising branded search with no new advertising or press is a strong hint that something upstream is working.

🤖 Your visibility in AI answers. Write down the five to ten questions your customers ask before they buy, and ask them in ChatGPT, Gemini and Perplexity each week. Note who gets named. This is the closest thing you have to a ranking report for AI, and it is free. This is exactly what our weekly AEO audits track for clients.

📊 Direct traffic and engagement together. Look at direct and organic visits alongside pages per visit and time on site. A rise in visitors who stay longer and look at more pages fits the pattern Similarweb found.

🧾 Sales, and where people say they heard about you. Add a simple "How did you hear about us?" question at checkout or on your enquiry form, and include "ChatGPT or another AI" as an option. It is low-tech, and it is the only way to hear about the journeys your analytics cannot see.

None of these on its own proves anything. Together, tracked over a couple of months, they tell you whether being recommended by AI is starting to pay off.

What this means for Australian businesses

Most of the research on AI search is American, and Australia tends to run a little behind on adoption. That cuts both ways. The AI share of your traffic may still look small, and this research suggests the true share is larger than it looks, because much of it arrives disguised as Google.

It also means the businesses that get named by AI now are building an advantage while their competitors are still deciding whether it matters. AI engines recommend businesses they can read and trust: clear answers to common questions, consistent business details everywhere online, reviews they can see, product and service information marked up properly, and a site that lets AI crawlers in. Every one of those can be fixed, and none of them depend on getting attribution perfect first.

So my advice is simple. Stop waiting for GA4 to show you a neat line that says "AI made you this much". It is not coming. Find out whether AI is naming your business when your customers ask, fix the reasons it is not, and watch your branded search and sales.

If you would like a starting point, our 7-Day AI Visibility Audit scores your site out of 100 against the ten signals AI uses to choose a business, then walks you through the fixes one day at a time. If your score does not go up in seven days, you get your $99 back.

FAQ

Does AI search actually send traffic to small business websites?

Yes, though much of it does not look like AI traffic. Similarweb found people recommended a brand by ChatGPT were 2.5 times more likely to visit that brand's site within seven days, and more than half of those visits came through search engines rather than a direct AI referral.

Why does ChatGPT traffic look so small in Google Analytics?

Because many people do not click the link in the AI answer. They search for the business on Google later, or type the address in directly. Your analytics records that final step, so the visit is credited to organic search or direct traffic instead of the AI tool that recommended you.

How can I tell if AI is recommending my business?

Ask the questions your customers ask, in ChatGPT, Gemini and Perplexity, and note which businesses are named. Do it weekly so you can see changes over time, and pair it with your branded search trend in Google Search Console.

Is AI traffic better quality than other traffic?

In the Similarweb research, AI-influenced visitors viewed nearly twice as many pages and spent roughly twice as long on site as other visitors. That suggests they arrive further along in their decision, although the research describes this as an association rather than proof of cause.

What is the cheapest way to measure AI's influence on sales?

Add a "How did you hear about us?" question to your checkout or enquiry form with an option for ChatGPT or another AI tool. Combined with branded search and weekly AI answer checks, it gives a small business a practical picture without paying for attribution software.

Sources

• Rob Tindula, NP Digital, What's Making AI Impact Difficult to Measure for Marketers?, 2 October 2026, citing Similarweb research on AI visibility's downstream impact.