# AI Crawling and Referral Traffic Are Different Signals

> Canonical source: [AI Crawling and Referral Traffic Are Different Signals](https://www.answer.cloud/pages/llms-like-chatgpt-claude-and-gemini-are-stingy-hosted/)

Read historical crawl-to-refer charts as request-to-referral ratios, not citation counts or guarantees for your website.

The historical charts below compare crawl requests with detectable referrals. They do not count citations in AI answers. See Cloudflare’s [2025 crawl-to-refer methodology](https://radar.cloudflare.com/year-in-review/2025#crawl-refer-ratio).

## Interpret the ratio correctly

The provider-level metric compares HTML crawl requests identified through user-agent headers with HTML referrals identified through referrer headers. A provider can have several agents. It is incorrect to assign the whole ratio to one bot or say a crawler “references your site” a fixed number of times before sending a visitor.

The values describe a historical network sample, not a promise for an individual website. Missing referrers and classification limits affect what can be observed.

Use your own request, referral, and conversion data alongside controlled answer tests. A high crawl count does not establish frequent citations or business value. Read the [metrics guide](https://www.answer.cloud/pages/llm-visibility-metrics-the-kpi-playbook-for-seo-and-marketing-leaders-hosted/).


![Article illustration](https://www.answer.cloud/assets/cms/article-23624-3efcd41ebfe0.png)


![Article illustration](https://www.answer.cloud/assets/cms/article-23624-de89211dc0d0.png)