AI search statistics, measured

The index

Live index counts, verified 2026-09-17

• Engagemii tracks more than 66 million companies and holds AI visibility scores on 54.7 million of them, the largest AI visibility index in existence.
• 42.1 million scored companies carry an industry classification across 22 industries.
• 3,507 US-listed public companies carry scores with ticker and SEC CIK attached, roughly 60% of the listed market.

Who is actually reading the web

Server-side crawler logs, 2026-05-22 through 2026-09-19

• AI crawlers made 137.4 million verified reads of Engagemii's network of business pages in under four months.
• Meta alone accounts for 71.4 million reads, roughly 52% of all logged AI reading, more than any other operator.
• Anthropic's ClaudeBot logged 33.6 million reads, about 3.5 times more than OpenAI's GPTBot at 9.7 million.
• AI reading grew 34x between May 2026 (2.0 million reads) and August 2026 (70.0 million reads).
• GPTBot reads at a nearly constant metered rate, around 166,000 reads per day for weeks at a stretch, while Meta arrives in bursts exceeding 3 million reads in a single day.
• Applebot's daily reads grew roughly 10x during September 2026; ClaudeBot's grew about 5x in the same month.
• PerplexityBot logged 218,000 reads in the same window, about 150x fewer than ClaudeBot.

The visibility bar

Score distribution across 42.1M categorized companies, 2026-09-18

• Only 4.4% of categorized scored companies reach an AEO score of 6 or higher.
• Only 0.07% reach a score of 8 or higher: about 30,000 companies out of 42.1 million.
• Technology has the deepest strong-scoring pool: 410,000 companies at 6+, yet still only 5.0% of the industry.
• Real Estate has the thinnest: 1.7% of 1.2 million scored companies reach 6+.

What predicts AI recommendations

Engagemii recommendation study, September 2026

• 78% of the brands ChatGPT recommended for real, unbranded questions scored 6 or higher, versus 7% of a random sample of scored companies.
• Recommended brands averaged a 6.5 AEO score; the random sample averaged 4.0.
• The relationship is graded: brands recommended once averaged 6.4, two to three times 6.6, four or more times 7.1.
• 84% of the domains ChatGPT recommended were already scored in the Engagemii index before the test ran.
• Put together: ChatGPT effectively recommends from the top 4.4% of the market, the same band the AEO score identifies.

Methodology in one paragraph

Index counts come from live database counts, not estimates. Crawler numbers are server-side log lines from pages Engagemii operates, with crawler identity verified against each operator’s published IP ranges; they describe reading of our network, which is large but not the whole web. Score-distribution figures cover the 42.1 million scored companies carrying an industry classification. The recommendation study methodology, including its predictive-not-causal caveat, is documented on the study page. Deeper crawler analysis lives in the crawler-reads study.