AI search is not a future trend. ChatGPT reached 200 million weekly active users in February 2025. Perplexity processes more than 100 million queries per month. Google's AI Overviews appear in roughly 30% of all search results. Users are already asking AI systems for business recommendations — and most businesses are not showing up.
The Core Numbers
- —91% of businesses audited by Zygur score below 40/100 on AI visibility
- —Average AI visibility score across audited sites: 23/100
- —Only 2–4 businesses per local market typically appear in AI recommendations
- —186M+ AI search queries per month across major platforms
- —Businesses with scores above 80/100 appear in AI recommendations up to 12× more often
- —73% of audited sites block at least one major AI crawler in their robots.txt
- —88% of sites have no llms.txt file
- —64% have no structured data AI systems can reliably parse
The Revenue at Stake
AI search is routing high-intent users to businesses that show up. In home services, AI platforms handle an estimated 52 million relevant queries per month. Only 2–4 providers per market are recommended consistently. In legal and finance, the number is 18 million monthly queries with similarly low competition at the top.
The businesses in AI recommendations today are not there because they paid for it. They are there because their sites were technically readable when AI platforms started indexing at scale. Most of them got lucky. The window to get there intentionally is still open.
Why Most Sites Fail the AI Visibility Audit
Robots.txt is the most common blocker
The most widespread issue is not a lack of content — it is that sites actively block AI crawlers. Many robots.txt files were written years ago with a Disallow: * rule, or with rules blocking unknown user agents by default. GPTBot, ClaudeBot, and PerplexityBot are relatively new user agent strings, so they get blocked alongside every other bot the site owner wanted to exclude.
Structured data is absent or broken
AI systems extract facts from structured data (JSON-LD) rather than reading prose. A page that says "We offer 24/7 plumbing services in Houston starting at $150" in plain text may or may not be understood correctly. The same information in a Service schema block with openingHoursSpecification and priceRange fields is unambiguous.
No llms.txt file
88% of sites have no llms.txt. This means AI crawlers have to guess at what the site is about from whatever pages they crawl. The result is inconsistent representation in AI answers — sometimes accurate, often not.
The Competitive Window
AI search visibility is still early-stage. Google's organic search competition took 20 years to mature. AI search competition is compressing that timeline, but the businesses that establish AI visibility now are building a durable position before paid AI search becomes the dominant channel.
Find out where you stand. Get your free AI visibility score at zygur.com in 60 seconds — no account required.