← Newsroom
Research

Only 216 of a million websites are ready to be recommended by AI

The Q3 2026 SAVI reading: 216 sites out of more than a million reach AI-Ready, not one has reached the top tier, and the highest score recorded anywhere in the corpus is 86.6 out of 100.

Only 216 of more than 1,000,000 websites analysed are ready to be found and recommended by AI assistants such as ChatGPT, Gemini, Claude and Perplexity, according to the SearchScore AI Visibility Index (SAVI) for the third quarter of 2026, published on 1 August. That is 0.022% of the sites studied, or roughly one in 4,600.

Not one site in the dataset reached the index's top tier, which requires a score of 90. The highest score recorded anywhere in the corpus is 86.6 out of 100.

The findings matter because of how buying decisions are shifting. A search engine returns a page of links and the customer chooses. An AI assistant returns one answer and names two or three businesses. Being absent from a page of links costs a business a click. Being absent from an assistant's answer costs it the entire conversation.

A fixed panel, so editions stay comparable

From Q3 the index is also tracked on a fixed panel of 102,873 domains, re-audited each wave on the same method so movement between editions reflects the web rather than a change in the sample. Wave one, dated 1 August 2026, sets the baseline at a weighted mean of 45.24.

The middle of the web is where the opportunity sits

The average website scores 43.2 out of 100 for AI visibility, and 63.8% score below 50, placing roughly two thirds of the web in the bottom half. The distribution matters more than the average. When someone asks an assistant for a recommendation it names two or three businesses, not ten, so where nearly every site looks the same to a machine, the few that stand out take far more than their share.

"Two thirds of the web scores below 50. That sounds like a crisis and it is closer to an opportunity, because your competitors are sitting in exactly the same place," said Ronnie Huss, founder of SearchScore.

How the index is built

The index is computed on a corpus of more than 1,000,000 websites analysed by SearchScore, each scored across more than 250 signals in 23 categories spanning AI visibility, SEO and conversion. A site is classed as AI-Ready at 80 points or above, and Verified at 90 or above. The corpus is stated with each quarterly edition so figures remain directly comparable between reports.

"Not one site in the dataset has reached the top tier," said Huss. "Nobody has won this yet, which means nobody is too late."

All figures are free to publish with attribution to SearchScore, and no permission is required. Custom cuts of the dataset for a specific sector, country or region are available on request, usually within a day. Every headline statistic, with its sample size and date, is published on the AI search statistics page, and the scoring method is documented at methodology.

Read the Q3 2026 report →

About SearchScore. SearchScore is a London-based AI search visibility platform and consultancy. It measures how easily AI answer engines and search engines find, trust and cite a website across AI search visibility (GEO), SEO and conversion (CRO), grounded in a benchmark of more than 1,000,000 audited sites. Its methodology is authored by founder Ronnie Huss.