A new round of data crunching has put a hard number on something marketers have been guessing at for two years: how much traffic AI chatbots actually send to websites. The answer, at least for ChatGPT relative to Google search, is 0.62 percent. That figure matters because an entire industry has sprung up selling tools to measure and improve something called AI visibility, often without much proof it moves the needle on actual business results.
What happened
Researchers compared several datasets that track how often brands and websites get mentioned, cited, or linked inside AI chatbot answers, then checked those figures against real referral traffic logs. The headline comparison: ChatGPT-driven visits to websites amount to roughly six-tenths of one percent of what Google's organic search sends. The analysis also looked at how well vendor dashboards that score a brand's AI visibility actually predict customer behavior like searches for a brand name or eventual purchases.
The short version is that many of these dashboards measure something real, called exposure, meaning how often a brand shows up in AI-generated answers. But exposure doesn't reliably predict recall, whether a customer later remembers the brand, or consideration, whether they actually consider buying from it. The data suggests a gap between a flashy visibility score and whether that score corresponds to any measurable business outcome.
This isn't the first sign that AI referral traffic is smaller than the marketing noise around it suggests. Server log studies published over the past year have consistently found that AI platforms, including ChatGPT, Perplexity, and Google's AI Overviews, send a sliver of the traffic that traditional search results do, even as those same platforms increasingly answer questions without the user ever clicking through to a website.
Why it matters
The pattern here is familiar. Whenever a new discovery channel emerges, whether it was banner ads, social media, or search engines in their early years, a wave of measurement vendors shows up before anyone agrees on a standard way to measure success. Social media had its "influence scores" that later proved weakly connected to sales. Early SEO had keyword density tools that measured activity, not outcomes. AI visibility tools appear to be following the same arc: real underlying activity, exists, but the scoring methodology is still catching up to what businesses actually need to know.
What this means for small businesses
If a vendor or consultant is pitching an AI visibility package, ask specifically what it's measuring and how that connects to leads or sales, not just mentions or citations in chatbot answers. A high visibility score with no tie to revenue is a vanity metric, and right now many available tools can't demonstrate that link convincingly.
Given that AI chatbots currently represent a fraction of a percent of the traffic that Google search still delivers, businesses with limited marketing budgets likely get more return from traditional SEO and local search listings than from chasing chatbot citations. That could change as AI search adoption grows, but it hasn't yet, based on the available referral data.
That said, ignoring AI visibility entirely isn't free of risk either. Being absent from AI-generated answers to questions about your industry could matter more in a year or two, even if it doesn't move traffic today. The practical move is modest: make sure basic business information is accurate and consistent across the web, since that's what AI systems pull from, without paying premium rates for unproven scoring dashboards.
What to watch
Track whether AI platforms start adding more clickable source links in their answers, which would signal growing referral potential, and watch for any vendor publishing a transparent, replicable methodology tying visibility scores to actual conversions rather than proprietary black-box numbers.
The bottom line
AI chatbots currently send a small fraction of the traffic that Google search does, and the tools that claim to measure and improve a brand's AI visibility have not yet proven a reliable connection to sales. Businesses evaluating these services should ask for evidence linking the score to revenue before spending on it.