The clearest signal in recent AI search data is audience overlap. In September, 93.7% of people who visited ChatGPT.com also visited Google.com, while 14.1% of Google.com visitors went to ChatGPT.com, according to Similarweb figures reported by Search Engine Watch. For marketers, that points to a measurement challenge: customers may use both services, while standard attribution systems assign credit to just one recorded visit.
These figures describe shared audiences, not a confirmed customer journey. They don’t establish that a person asked ChatGPT for a recommendation and then searched Google for a brand. The data doesn’t show visit order, search frequency, queries, purchases or activity inside either service’s app. Treating it as proof that one channel sends customers to the other would stretch the evidence.
Audience overlap changes how reach should be counted
Marketing plans often assign distinct roles to search, social media, referrals and paid campaigns. AI assistants make those boundaries less reliable. A prospective customer might use an AI service to compare products, search a company name on Google, watch a review and then navigate directly to the company’s site. The reported overlap doesn’t verify that sequence, but it does show why marketers shouldn’t assume AI and Google audiences are separate groups.
That distinction affects decisions about reach and spending. If a company adds the full audience of an AI platform to its Google audience, it may count some people twice. If it treats a rise in AI use as evidence that customers have stopped using Google, it may draw the opposite mistake. The data supports a narrower conclusion: many visitors use both websites within the same month.
My read is that marketers need to plan for connected discovery rather than a winner-takes-all shift between platforms. Customers can add a new research tool without dropping familiar search habits. The practical question is whether a brand remains visible and credible as people move between sources.
Last-click reports leave part of the journey out
A customer might first encounter a company in an AI-generated answer and later search its name on Google. The resulting website session could be credited to organic search, even if the earlier exposure helped create the interest. An AI interaction might also shape a decision without producing a click or a visit that the company can track. The Similarweb data can’t distinguish among these possibilities.
Last-click attribution records the source associated with a measured session. It generally can’t identify every earlier interaction that influenced a customer to seek out a brand. That makes it a limited guide to AI’s contribution, especially when an assistant provides information without sending a visitor directly to a company’s site.
Google’s Search Console generative AI performance report offers one additional view. It shows impressions for pages appearing in Google’s AI search features, including AI Overviews and AI Mode. Those impressions help companies monitor visibility within Google’s AI products, but they don’t establish incremental traffic or revenue. The report also doesn’t cover ChatGPT or other AI platforms.
Pair visibility measures with business outcomes
Companies can make the data more useful by tracking several measures together. Traditional search clicks show visits from search results. Branded searches can indicate whether more people are actively looking for the company. Lead quality and sales connect discovery to commercial outcomes, though they don’t identify which channel caused the result by themselves.
Alongside those measures, marketers can monitor which pages appear in Google’s AI features and whether AI services cite the company in a consistent set of relevant prompts. Asking customers where they first heard about the business, through a lead form or post-purchase survey, adds another piece of evidence. Each measure has limits; viewed together and over time, they can show patterns a last-click report misses.
Teams should be careful when interpreting changes across dashboards. A drop in search clicks alongside higher AI visibility doesn’t automatically mean that AI has replaced Google traffic, or that the new visibility is producing customers. Compare branded searches, direct visits, lead quality and conversions across periods, then consider what customers say influenced their decisions. A single month of traffic data is a weak basis for reallocating budgets.
Build a discovery plan across platforms
The figures don’t prescribe a split between Google and AI. They do argue against planning as though every platform reaches an entirely different customer. Search work still matters: companies need accurate pages explaining their products, pricing and value, as well as a clear view of which questions bring qualified prospects to their sites.
The same attention should extend to AI services. Marketers can check whether those systems describe the company accurately, draw on useful sources and include it when customers compare relevant options. That work is easier to evaluate when search, content and analytics teams share information instead of reporting platform performance in isolation.
The 93.7% overlap figure doesn’t show that Google is insulated from competition or that AI search is reducing existing traffic. It measures visitors to two websites in September, with no evidence about sequence or purchase behavior. Its value is as a reminder that platform audiences can overlap and that familiar attribution methods may miss earlier discovery.
For business leaders, the useful takeaway is to track presence across search and AI alongside demand and sales indicators. Customers may encounter a company in an AI answer, then look it up on Google before acting. Marketing reports that account for both moments will give a more grounded view of how discovery is changing.
