For all the promises about AI-powered search delivering higher-quality traffic, the data is painting a more nuanced picture. A new study from Ahrefs, analysing nearly 82,000 websites between May and June 2025, reveals a clear divergence in user behaviour between traditional search traffic and visits from AI platforms like ChatGPT, Perplexity, and Copilot. The findings raise serious questions about the real-world value of AI traffic and its impact on engagement, experience, and conversions.

AI Users Engage Less Across the Board

Marketers have long focused on growing “high-intent” traffic, where users take actions that indicate deeper engagement, clicking through multiple pages, staying longer, and converting. Google has repeatedly said AI-powered results lead to better, more qualified visits. But when you measure actual on-site behaviour, AI traffic often falls short.

According to the Ahrefs data:

• AI users bounce at a rate of 67.8%, compared to 63.7% from search visitors and 62.4% for all traffic sources.

• They view 4 pages per visit on average. Search users clock in at 5.2, while overall traffic hits 5.5.

• Session depth is also lower. AI visitors browse 2.27 pages per session minute, while search delivers 2.79 and all traffic averages 2.99.

• Session duration is the only outlier, with AI visitors staying 8 seconds longer (86 seconds) than either search or average users.

In short, AI traffic appears more decisive but less engaged. Visitors spend slightly more time per session but touch fewer parts of the site. They arrive, they scan, and they leave.

Does That Make AI Traffic Worse?

Not necessarily. What we’re seeing is not that AI traffic is low quality, it’s that it behaves differently. Users are coming in with more focused queries, often having already filtered their options via AI chat interfaces. As a result, they don’t need to browse multiple pages or dig deeply into content. In some contexts, that may actually be desirable.

If someone visits a product page and converts in one click, that’s a success, even if they never browse another page. The higher bounce rate might reflect more efficient decision-making, rather than disinterest or dissatisfaction.

Yet, there’s a danger in assuming the best-case scenario. For many publishers and brands, especially those dependent on multi-page visits for ad revenue or lead nurturing, this “one-and-done” behaviour is problematic.

Conversion Rates Tell a Different Story

One potential bright spot? Conversions. The same Ahrefs study suggests AI traffic might be outperforming search in this department.

Several companies reported higher conversion rates from AI channels:

• Ahrefs observed a 12% higher conversion rate from AI platforms compared to organic search.

• Vercel claims a 10% CVR from AI sources.

• Tally attributes over $1M in ARR growth to traffic from AI channels.

These early indicators support the idea that AI visitors may be more qualified, even if they are less exploratory. Users landing from platforms like ChatGPT often arrive with a specific outcome in mind. They may be closer to a decision point and more ready to act.

However, the conversion data is still anecdotal. Conversion tracking across AI platforms remains limited. Google, for example, still does not let marketers distinguish between traditional search and AI Overview traffic in Search Console or Analytics. Without this visibility, it’s impossible to validate claims of higher traffic quality from AI features in Google Search.

Google’s Narrative Needs More Transparency

Google’s messaging around AI traffic is consistent: fewer but better clicks. A more qualified user. A more meaningful visit. But these are difficult claims to evaluate without data.

Elizabeth Reid, Head of Google Search, recently said that AI Overviews bring down friction, increase search frequency, and drive better clicks for websites. An unnamed Google exec added that AI-generated search results will ultimately lead to more time on site and better engagement. The problem is, Google doesn’t let us measure any of this independently.

So, we turn to external data sources like Ahrefs. Based on their findings, AI traffic is less exploratory, more transactional, and structurally different. Whether that’s “better” or not depends on your business model.

What Marketers Should Do Next

If your website relies on high page views, ad impressions, or discovery-based engagement, AI traffic may challenge your assumptions. You might see fewer pages per visit and higher bounce rates. On the flip side, if you’re focused on conversions or specific user actions, AI traffic might prove more efficient.

Here’s how to adapt:

• Re-evaluate engagement metrics. Bounce rate and pages per session may no longer reflect user intent accurately, especially for AI traffic.

• Start tracking conversions rigorously. Even if you can’t isolate AI sources fully, get granular with performance by referral domain or landing page.

• Optimise for landing-page success. If AI traffic arrives with strong intent, ensure your entry pages are conversion-ready with clear CTAs and focused messaging.

• Prepare for AI search visibility shifts. Google’s AI Overviews and chat-style answers are likely to change what drives impressions and clicks. Expect fluctuations in organic traffic patterns.

Bottom Line

AI-driven visits aren’t necessarily better or worse than traditional search traffic. They are simply different. Marketers should treat them as such and resist the temptation to use outdated metrics as the sole benchmarks of success. Engagement might be lower, but conversion potential could be higher.

Until platforms like Google and Bing offer proper segmentation of AI and traditional traffic in their tools, we’ll remain reliant on third-party studies and anecdotal evidence. But that doesn’t mean we can’t start adapting now.

Measure differently. Optimise smarter. And keep an eye on how AI user behaviour evolves.

It’s not about quantity or quality. It’s about fit.