AI referral traffic is still small enough to mislead both optimists and skeptics. Ahrefs’ 2026 analysis of 74,752 sites found all AI chatbots combined sent 3.5 million visitors in March—0.28% of total traffic in its dataset. That is not a reason to abandon search. It is a reason to measure a different kind of intent.
Separate humans from bots
AI crawlers fetch pages; AI referral visitors clicked a cited answer. These are different events. Track them separately in analytics, then compare engaged sessions, email capture, demos or purchases—not just visits.
Build pages people can cite and use
Pages with a specific job tend to travel better than generic commentary: honest comparisons, step-by-step guides, product pages, free tools and updated research. The target is not to “game AI.” It is to make one page more useful when someone has a precise question.
A simple measurement setup
- Create a referral group for ChatGPT, Perplexity, Claude, Gemini and Copilot when analytics makes that possible.
- Track one meaningful event: email signup, saved result, trial activation or qualified contact.
- Review landing pages monthly: which page earns citations, which earns clicks, and which earns outcomes?
- Update the pages whose facts or screenshots have changed.
Freshness is a product decision
AI systems and people both prefer reliable answers. Add visible update dates only when you actually review the facts. Replace stale screenshots. Explain limits. Link to primary sources. These choices are more durable than building pages for every possible wording of a question.
Source note
Ahrefs’ AI chatbot traffic analysis is the data source; its figures reflect the studied dataset, not the entire web.
