AI Search and Google Audience Overlap: What the 93.7% Figure Means for Marketers

For marketers, the clearest signal in Similarweb’s September audience data is that ChatGPT and Google visitors overlap. The figure complicates the idea that AI search has created a separate customer pool, but it doesn’t show whether AI tools are taking traffic from Google, creating demand, or contributing to sales.

Similarweb reported that 93.7% of ChatGPT.com visitors also visited Google.com during the month. In the other direction, 14.1% of Google.com visitors also visited ChatGPT.com, according to Search Engine Watch’s report. Those numbers describe shared visitors across two websites. They don’t establish which service came first, what people searched for, or whether any visit led to a purchase.

That distinction matters for planning. If a company treats Google and AI platforms as wholly separate audiences, it could overstate its reach, misread channel performance, or assign credit to the last click while missing earlier discovery. The data makes a case for measuring customer journeys with more care, not for declaring one channel the winner.

Shared visitors, unknown journeys

The overlap figures answer a narrow question: how many people visited both sites during the measured month? They don’t reveal a sequence such as asking ChatGPT to compare vendors, searching for a brand on Google, then visiting its website. That path is plausible, but the audience data doesn’t confirm it.

Nor does the measure capture every interaction with either service. A person may use a mobile app rather than visit ChatGPT.com or Google.com in a browser. The reported overlap therefore shouldn’t be treated as a complete record of how customers use search or AI tools.

Still, the results are useful as a warning against tidy channel assumptions. A person can use several sources while researching a purchase. They might start with an AI search answer, check a company through traditional search, review a creator’s opinion, and then go straight to the company’s site. Each service may play a different role, and an analytics report may only capture the visit that produced a trackable click.

That makes audience overlap a planning consideration, not proof of influence. It also makes it risky to add reach figures from different platforms together as if each represented a distinct group of potential customers. Where users appear in more than one audience, separate platform totals can count the same person twice.

Attribution can miss earlier discovery

Last-click reporting is particularly limited when customers move between platforms. A visitor who first encounters a company in an AI response might later search its name on Google. In an analytics report, the later organic visit could receive the credit, while the earlier exposure remains unrecorded. The reverse is possible, too: an AI interaction may occur after a customer has already found the brand elsewhere.

Neither scenario can be verified from Similarweb’s overlap statistic. The figure does not connect an individual’s browsing sequence to a search, lead, or transaction. It provides no evidence about whether an AI interaction changed purchasing intent or generated incremental demand.

Google has added one source of visibility data through its Search Console generative AI performance report. It shows impressions for pages that appear in Google’s AI search features, including AI Overviews and AI Mode. That can help site owners understand whether their pages appear in those features, but it doesn’t establish that the visibility produced additional traffic or revenue. It also doesn’t measure appearances on ChatGPT or other AI platforms.

Marketers should keep those measures distinct. Traditional search clicks can show that a visit came from Google; branded search volume can indicate that more people are looking for a company by name; qualified leads and sales show outcomes closer to commercial value. AI feature impressions describe visibility in those features, while brand citations in relevant AI answers can offer another signal to monitor. None alone proves that AI drove a sale.

Customer feedback can add context that platform reporting lacks. A brief question on a lead form or post-purchase survey—“Where did you first hear about us?”—won’t give a complete attribution model. But answers collected over time can help identify discovery sources that don’t appear in referral data.

Coordinate search and AI measurement

The practical response isn’t to divide marketing resources evenly between Google and AI tools based on the overlap percentage. The figure contains no budget guidance. It does suggest that teams should avoid planning around the assumption that each platform reaches a different person.

Companies can start with the material customers rely on to assess a business: clear, current pages describing products, pricing, and value. Existing SEO work can help customers find and evaluate those pages. Teams can also review which questions bring qualified prospects to the site, then check whether AI answer engines describe the business accurately and cite useful sources when customers compare options.

This work needs a shared view of performance. If search traffic declines while visibility in AI features rises, that combination alone doesn’t establish that AI has replaced lost Google visits. Teams should compare branded searches, direct visits, lead quality, and conversion trends over time. They can then look for consistent changes across several measures and ask customers what influenced their decisions.

One month’s audience data is a thin basis for strategic conclusions. The September finding doesn’t establish that Google is insulated from competition, and it doesn’t show that AI search is reducing existing traffic. Its value is more specific: many people who visit ChatGPT.com also visit Google.com, while a smaller share of Google.com visitors also visit ChatGPT.com.

For businesses, that is a reason to account for overlapping audiences and imperfect attribution. Customers may encounter a company through AI, use Google to check it, and convert through a later visit that gets credit for only one step. Better decisions will come from tracking visibility, branded demand, lead quality, and sales together—and treating the journey as a sequence of interactions that no single channel report fully captures.

As AI search develops, the useful question for marketing teams is how each source contributes to discovery and evaluation. The overlap numbers establish that the audiences intersect; the commercial effect will need to be assessed through broader, ongoing measurement.