When potential customers ask questions related to your industry in ChatGPT, Claude, Perplexity, or Google AI Search, the AI Crawlers/Bots sent by these AI services visit your public website to read your content.
Because AI bots do not execute front-end JavaScript tracking codes like human visitors, Kanorio has built a dedicated server-side tracking pipeline. This displays all known AI and search crawler access logs in the "Analytics › AI Bots" tab—ensuring your human traffic data remains accurate while giving you clear visibility into which content is most frequently cited by AI!
Looking to track "human" clicks from ChatGPT or Gemini, or adjust bot blocking rules?
- This page covers "AI automated crawlers visiting your website." If you want to see statistics on "human visitors clicking citation links in ChatGPT, Google Gemini, or Claude responses to reach your site," please go to "Analytics › Overview" and switch the traffic source to "AI Assistant" (see ).
- If you want to learn how to improve your AI search citation rate (AEO/GEO) or set up "Block Training, Allow Retrieval" strategies, please refer to .
Open "Analytics › AI Bots". The four cards at the top display the totals for the selected period and the percentage change compared to the previous period:
| Metric Card | Meaning | Key Takeaways |
|---|---|---|
| Total Visits | Total requests from known AI and search engine bots reading your public website pages (excluding human traffic). | Monitor whether your website content is being consistently indexed and updated by major AI and search engines. |
| Blocked | The number of requests from bots you have set to block in your "Blocking Rules," or requests rejected by the website with a 403 Forbidden status. | Click this card to open the "AI Bot Blocking Rules" panel. Blocked requests are still recorded here, proving that your firewall is successfully protecting your content! |
| IP Verified | The number of requests where the source IP matches the official server IP ranges published by the AI operator. | These are authentic bots from official sources like OpenAI, Google, Anthropic, Perplexity, Apple, Meta, and Microsoft. |
| Suspected Spoofing | The number of requests where the User-Agent claims to be a well-known AI bot, but the source IP does not match the official published ranges. | Commonly seen when third-party crawlers impersonate GPTBot or Googlebot to scrape your pages. |
The rankings and bar charts in the report categorize 35 types of bots from 18 operators—including OpenAI, Anthropic (Claude), Google, Perplexity, Meta, Apple, Microsoft (Bing), Amazon, xAI, Mistral, Cohere, DuckDuckGo, ByteDance, You.com, Common Crawl, Diffbot, AI2, and iAsk—into three categories based on their public use cases:
ChatGPT-User, OAI-SearchBot, PerplexityBot, Claude-SearchBot, and DuckAssistBot. When a user asks a question in an AI chat interface, these bots read your page in real-time and include your site link. We recommend keeping these "Allowed" to ensure you receive free AI search exposure.GPTBot, ClaudeBot, Google-Extended, Meta-ExternalAgent, and CCBot. These bots read content to train the next generation of AI models and may not include your URL in their responses. If you prioritize original content copyright, you can set training bots to "Blocked" at any time.Googlebot, Bingbot, and Applebot. These bots ensure your site appears in Google and Bing search results. (This is vital for site visibility; the system protects these by default. To disable full-site indexing, please go to ).Any program on the internet can change its User-Agent to GPTBot. To prevent fake crawlers from skewing your data, Kanorio automatically verifies them against official public IP lists (CIDR and ASN) on the server side:
| Verification Status | System Logic | How to Interpret |
|---|---|---|
| Verified | The request's connection IP falls within the official server IP ranges published by OpenAI, Anthropic (Claude), Google, Perplexity, Apple, Bing, Meta, Amazon, etc. | You can be 100% confident that these are authentic official visits. |
| Spoofed | The request claims to be a well-known bot with a public IP list, but the connection IP is not within the official range. | This indicates an external third-party program is impersonating a bot. If it occurs sporadically, you can ignore it; if you notice unusually high scraping activity, you can block that type or contact support. |
| Unverifiable | The company behind the bot (such as some AI startups or services without public IP lists) has not yet published fixed IP ranges for verification. | "Unverifiable" does not mean malicious or fake; it simply means the operator has not yet made their official IP list public for global verification. If you use your own headless frontend to report crawler visits, Kanorio receives the IP of your frontend server, and these visits will also be classified here. |
Seeing them means your site has successfully entered the AI's retrieval radar and is not blocked by technical barriers! Whether an AI ultimately decides to cite your page in a specific answer depends on whether your content directly answers the user's question and if your structure is clear. To increase the probability of being cited by AI, please refer to the writing suggestions in AI Search Optimization and Bot Management.
For newly published sites, AI crawlers and search engines need time to discover your URL through external links, Sitemaps, or IndexNow. We recommend completing the "SEO › Overview" Google Search Console one-click registration. You will typically see bots visiting within a few days to a few weeks.
Dashboard totals, bot rankings, and page-level intent summaries (AI Search/Training/Search Engine reads) are stored permanently. You can select any historical period (up to 90 days per query) in the top right corner to view and compare. Page-level intent summaries are a newer report, with data available from 90 days before the feature launch; earlier dates only show totals and rankings. If querying via AI Assistant (MCP) or Management API, you can query up to 10 years (3,650 days) of totals and page-level intent; details on "which bot read which page how many times" are only retained for the last 90 days.