The AI Visibility Index: how 10 leading sites score with AI search
An audit of ten major public homepages reveals that while most have adopted basic AI visibility protocols, significant gaps in structured data and crawler access persist, resulting in an average visibility score of 62.3 out of 100.
Key findings
- The aggregate AI-visibility score across the ten sites is 62.3/100, with Stripe.com leading at 86.0 and OpenAI.com trailing at 22.0.
- 70% of the audited sites (7/10) have implemented an llms.txt file, indicating widespread adoption of this protocol for guiding large language models.
- Only 50% of the sites (5/10) ship schema.org structured data, suggesting that half of these high-traffic domains are not fully optimizing their markup for AI parsing.
- Two sites actively block at least one AI crawler, creating potential blind spots in how their content is ingested by generative models.
- There is a notable divergence in performance: sites with both llms.txt and schema.org generally score higher, while the absence of both correlates with lower visibility scores.
The ranking
| # | Site | Score | llms.txt | Schema |
|---|---|---|---|---|
| 1 | stripe.com | 86 A | yes | yes |
| 2 | www.cloudflare.com | 86 A | yes | yes |
| 3 | www.shopify.com | 78 B | yes | yes |
| 4 | github.com | 69 C | yes | no |
| 5 | linear.app | 64 C | yes | no |
| 6 | www.notion.com | 61 C | yes | no |
| 7 | vercel.com | 60 C | yes | yes |
| 8 | www.figma.com | 56 C | no | yes |
| 9 | www.anthropic.com | 42 D | no | no |
| 10 | openai.com | 22 F | no | no |
What the data means
The data suggests that while the industry has largely standardized on the llms.txt file as a primary signal for AI crawlers, this alone is insufficient for high visibility. Seven out of ten sites have adopted llms.txt, yet only five achieved scores above the median. This indicates that the presence of the file is a baseline requirement rather than a differentiator. Stripe.com and Cloudflare.com, which scored above 85, exemplify the optimal configuration: they possess llms.txt, ship schema.org data, and likely offer robust access to their content, resulting in superior AI visibility.
Takeaways
- Implementing llms.txt is now a standard practice for major tech brands, but it should be viewed as a foundational step rather than a complete solution for AI visibility.
- Schema.org structured data remains underutilized, with half of the top-tier sites lacking it, which may limit the depth and accuracy of AI-generated responses derived from their content.
- High traffic volume does not guarantee high AI visibility; OpenAI.com’s low score of 22.0 demonstrates that even the creators of leading AI models can have suboptimal visibility configurations on their own homepages.
- Organizations seeking to improve their presence in AI search results should prioritize the combination of llms.txt and schema.org markup to maximize both accessibility and data richness.
- Crawler blocking strategies should be reviewed to ensure that key AI bots are not inadvertently excluded, as this can significantly impact how content is indexed and cited by generative models.
Run a free Heron audit