AI Visibility Index: 94.8% of Businesses Remain Invisible to AI Assistants
New data reveals a massive gap between AI crawler access and actual citations in AI-generated answers.
The vast majority of businesses are failing to surface in AI-generated responses, even when they leave their digital doors open to crawlers. A new report shows that accessibility does not equal visibility in the era of generative search.
According to the second edition of the AI Visibility Index released by Website Auditor in August 2026, 94.8% of audited sites never appeared in an AI answer for buying-intent queries across ChatGPT, Claude, Gemini, and Perplexity. Despite this invisibility, only 8.9% of the audited sites actually block at least one AI crawler in their robots.txt files. Among the tested assistants, Gemini demonstrated the widest reach, naming audited businesses in 2.9% of its responses.
The Retrieval Gap
As traditional search is replaced by Generative Engine Optimization (GEO), businesses are increasingly focused on "Answer Presence" and "Machine Readability." However, the data suggests a critical misunderstanding of how AI agents operate. The study found a significant disparity in blocking behavior: 38 audited sites block GPTBot, which is used for training, while only four block OAI-SearchBot, which is used for real-time retrieval.
Website Auditor notes that blocking retrieval agents, such as OAI-SearchBot and Perplexity-User, effectively makes a site "uncitable." This carries a far more severe consequence for immediate business visibility than blocking training crawlers, which only affects the model's long-term knowledge base.
The Role of Structured Data
Mere accessibility is insufficient for securing a citation. The index indicates that AI citation is driven by high-density, specific content and structured data rather than simple crawlability. A lack of technical signals is evident in the results, with only 19.3% of audited sites utilizing LocalBusiness schema—the least-adopted machine-readability signal measured in the study.
This suggests that while most sites are technically reachable, they lack the structured markers necessary for AI assistants to confidently identify and recommend them to users during the buying process.
Future Outlook
For site owners, the primary takeaway is that misconfiguring robots.txt to block retrieval agents can actively eliminate any chance of appearing in real-time AI responses. As AI-powered search continues to evolve, the focus for businesses will likely shift from basic crawler permissions to the aggressive implementation of schema.org markup and high-density content to bridge the visibility gap.