Ahrefs says AI chatbots sent 3.5 million visits, a small but measurable channel

Why AI chatbot traffic matters now

AI chatbot referrals are still a sliver of the web, but Ahrefs has finally put a hard number on the opportunity: 3.5 million visitors in March 2026 across 74,752 websites, equal to 0.28% of total web traffic. That is not a replacement for search, and it is nowhere near Google’s scale, but it is no longer a vague side effect of AI usage. It is a measurable channel with enough volume to justify reporting, optimization, and client-side planning.

The practical point for agencies is simple: traffic from chatbots behaves differently from ordinary organic search. Ahrefs says this is human traffic, meaning a user clicked out of a chatbot response after the AI cited a page. That is distinct from bot crawl traffic from systems such as GPTBot and ClaudeBot, which should not be confused with actual visitors.

What the numbers say about scale

The comparison with Google makes the channel’s current size clear. In the same March 2026 dataset, Google sent 345.2 million visits, or 28.12% of total web traffic. Chatbots are still tiny next to that, and ChatGPT was the largest single contributor among AI sources, with Perplexity, Gemini, and Claude trailing behind.

Even so, the spread across 74,752 websites matters. This is not a niche effect limited to a handful of brand names or viral articles. It is broad enough that agencies can start tracking it at the account level, especially for clients with strong informational content, product comparisons, or research-heavy buyer journeys.

Why this is more than a curiosity for agencies

The biggest mistake is treating chatbot referrals as anecdotal noise. If the traffic can be measured, it can be attributed, and if it can be attributed, it can be reported as incremental value. That changes the conversation with clients, especially when traditional search performance is flat or declining.

Ahrefs’ own research strengthens the commercial case. The company says AI search visitors convert at a 23x higher rate than traditional organic search visitors for Ahrefs. Even if most brands will not see that exact ratio, the message is hard to miss: a small stream of highly qualified clicks can punch above its weight when the visitor is arriving after an AI system has already framed the recommendation.

How the channel has grown in Ahrefs’ broader research

The March 2026 snapshot fits a longer pattern. In an earlier Ahrefs study of about 3,000 sites, 63% received at least one visit from an AI source. In another study across roughly 35,000 websites, AI accounted for just 0.1% of total referral traffic. Those figures show a channel that started tiny, spread widely, and is now growing into something agencies can no longer ignore.

Ahrefs later updated that picture again, saying AI traffic had grown about 9.7x year over year across 81,947 sites, while average search traffic fell about 21%. That combination is the real strategic warning. AI referral traffic is not yet large, but it is growing fast enough to matter while traditional search traffic is getting harder to win.

The push and pull between AI platforms and publishers

The broader debate around AI visibility is not just about mentions inside chat interfaces. It is about whether those mentions turn into clicks that reach the open web. Google pushed back on the idea that AI search weakens the web, saying in August 2025 that AI in Search is driving more queries and higher-quality clicks, while still sending billions of clicks to the web every day.

Publisher-side data tells a more cautious story. Similarweb data reported by Digiday showed ChatGPT referral traffic to 250 news and media sites reached 243.8 million visits in April 2025, up 98% from January 2025. Digiday also reported that 83% of OpenAI’s generative AI outbound traffic went to news and media sites in that month. That is a reminder that AI intermediaries can move real audience flows, but the gains are uneven and still evolving.

How agencies should track AI chatbot traffic

To make this channel useful, it needs to be handled like a distinct acquisition source, not folded into generic referral noise. The first step is building clean reporting that separates human chatbot visits from bot crawl activity. If GPTBot, ClaudeBot, and similar crawlers are mixed with actual users, the story gets distorted immediately.

A practical agency workflow should include:

• A dedicated channel grouping for AI chatbot referrals in analytics
• Clear source rules for known assistants such as ChatGPT, Perplexity, Gemini, and Claude
• Landing-page tracking to see which pages are earning citations and clicks
• Conversion tracking tied to forms, demo requests, newsletter signups, or ecommerce revenue
• Monthly trend reporting so clients can see whether citations are producing durable traffic, not one-off spikes

The goal is to show whether AI referrals are merely present or actually profitable. A traffic line that is only 0.28% of web traffic can still be useful if it over-indexes on leads or sales.

How to grow the channel before it gets crowded

Growth comes from being easier for AI systems to cite and easier for humans to trust once they click through. That means content should answer specific questions clearly, use strong page structure, and cover the kind of factual, comparison-driven topics AI tools tend to pull into responses. Pages that are already strong in search often have the right raw material, but they may need sharper summaries, better topical clarity, and more explicit evidence.

Agencies should also think in terms of citation-worthy assets. Product pages, definitional guides, pricing explainers, and industry analysis tend to be the kinds of pages that can be surfaced in a chatbot answer and then clicked. The channel grows when a page is both legible to an AI system and useful enough that a human wants the next step after reading the summary.

The strategic takeaway

AI chatbot traffic is not replacing search, and Ahrefs’ own numbers prove that Google still dominates by a wide margin. But the channel is real, measurable, and already showing signs of commercial value through high-intent clicks and strong conversion potential. Agencies that start tracking it now will have a cleaner story for clients, better attribution, and a head start before AI referrals become as crowded and competitive as search itself.