I keep hearing a version of the same question from business leaders: Is AI search taking customers away from Google?

A new Similarweb audience-overlap figure offers a useful answer, with an important limit. In September, 93.7% of ChatGPT.com’s visitors also visited Google.com. The overlap ran the other way, too: 14.1% of Google.com’s visitors also visited ChatGPT.com, according to Search Engine Watch’s report.

That is not proof that someone asked ChatGPT about a product, then searched for it on Google. The data counts people who visited both websites during the month. It doesn’t show which one they used first, how often they searched, what they searched for or whether they bought anything. It also doesn’t capture the full use of either service through apps.

Those limits matter. But they don’t make the finding irrelevant. They tell us not to mistake channel overlap for channel substitution.

One customer, more than one discovery path

For years, marketers have organized discovery into tidy channels: search, social, referrals, paid media. AI assistants make that division harder to defend. A person might ask an AI tool to compare options, search a brand name in Google, watch a creator review, then visit a company’s site directly. The journey may cross several systems before the business gets a measurable click.

The Similarweb figure doesn’t prove that sequence. It does show why we shouldn’t assume that a visitor counted in an AI audience is a new person who has never used Google. Treating those groups as separate pools can lead to double-counting, inflated reach estimates and bad budget decisions.

I saw a version of this mistake when search was still new to many companies. Some treated SEO as a separate experiment instead of infrastructure for demand that already existed. The risk now is the reverse: treating AI discovery as a replacement for the search habits customers still have.

The better question isn’t which platform wins. It’s whether your brand is present, clear and credible when a customer moves between them.

Don’t confuse visibility with a sale

The overlap stat also exposes a measurement problem. A customer might encounter a brand in an AI answer and later search for it by name. That later visit could show up as Google organic traffic, even if the earlier discovery helped create the demand. Or the AI interaction might never produce a visit at all. The audience-overlap data can’t tell us which happened.

That is why last-click reporting is a poor referee for this shift. A referral source can tell you where a session came from. It usually can’t tell you which earlier interactions made the brand worth looking up.

Google has begun giving site owners a clearer view of one part of the picture. Its Search Console generative AI performance report shows impressions for pages appearing in Google’s AI search features, including AI Overviews and AI Mode. That is useful visibility data, not proof of incremental traffic or revenue, and it doesn’t cover ChatGPT or other AI platforms.

Businesses should keep the measures separate. Track traditional search clicks, branded searches, qualified leads and sales. Alongside them, monitor which pages appear in Google’s AI features and whether your brand is cited in a consistent set of relevant AI prompts. Add a simple question to lead forms or post-purchase surveys: “Where did you first hear about us?” None of those signals proves, on its own, that AI caused a sale. Together, they can reveal patterns that last-click reports miss.

Build one discovery plan, not two channel silos

The response isn’t to split the marketing budget evenly between Google and AI. It’s to stop planning as if every platform reaches a different customer.

Keep investing in the parts of search that help customers find and evaluate you. Make sure the pages that explain your products, pricing and value are current. Understand which questions bring qualified prospects to your site. Then extend the same discipline to AI discovery: Check whether answer engines describe your business accurately, cite useful sources and include you when customers compare options.

Most importantly, connect the teams and dashboards. If SEO reports a decline in clicks while AI visibility rises, don’t declare one channel the winner and the other the loser. Look at branded search, lead quality, direct visits and conversion trends. Ask customers what influenced them. Compare the pattern over time, not just across one month’s traffic report.

I don’t read the 93.7% figure as evidence that Google is safe or that AI search is harmless to existing traffic. It doesn’t answer either question. I read it as a reason to be more precise: Customers can add a new way to discover a business without abandoning the old one.

Your customers may meet you in an AI answer and Google you before they act. The companies that understand both moments will make better decisions than the ones still asking which channel gets credit for the whole journey.

I keep hearing a version of the same question from business leaders: Is AI search taking customers away from Google?

A new Similarweb audience-overlap figure offers a useful answer, with an important limit. In September, 93.7% of ChatGPT.com’s visitors also visited Google.com. The overlap ran the other way, too: 14.1% of Google.com’s visitors also visited ChatGPT.com, according to Search Engine Watch’s report.

That is not proof that someone asked ChatGPT about a product, then searched for it on Google. The data counts people who visited both websites during the month. It doesn’t show which one they used first, how often they searched, what they searched for or whether they bought anything. It also doesn’t capture the full use of either service through apps.