See every agent that visits your site
From Googlebot to GPTBot: search crawlers, AI assistants, social preview bots, SEO tools, and scrapers. BotSights names them, verifies they are who they claim to be, and connects their visits to what matters: rankings, citations, shares, and customers.
What is agent analytics?
Agent analytics is the measurement of your automated visitors: the search crawlers, AI assistants, social preview bots, SEO tools, and scrapers that read websites on behalf of platforms and their users. Where web analytics counts human sessions, agent analytics covers the other half of your traffic and answers: which agents visit your site, which pages do they read, how often, and on whose behalf?
It matters because a growing share of your audience never loads your page in a browser. They ask an assistant, the assistant visits for them, and your analytics see nothing. Automated traffic already outweighs human traffic: Imperva measured 51% of all web traffic as bots in 2024. The agent visit is the new pageview: it is where citations, recommendations, and AI-referred customers begin.
Every kind of agent, one dashboard
BotSights identifies 1,900+ known bots and agents, categorized by what they do for you, with AI as the fastest-growing slice. The database grows as new agents appear.
AI Agents
GPTBot, ClaudeBot, ChatGPT-User, PerplexityBot
The fastest-growing category: assistants fetching pages for users, AI search crawlers building answer indexes, and training scrapers. Every visit here shapes whether AI cites you.
Search Engines
Googlebot, Bingbot, DuckDuckBot
Your indexing health, measured. Crawl frequency per page shows what search engines value, and a crawl drop warns you weeks before rankings move.
Social Preview Bots
WhatsApp, facebookexternalhit, LinkedInBot
Every link shared in a chat or feed triggers a preview fetch. These are shares Google Analytics never sees, most of them in private messages.
SEO & Marketing Tools
AhrefsBot, SemrushBot, rogerbot
Who is auditing your site, and for whom? Tool crawlers reveal competitor research and partner activity on your domain.
Scrapers & Data Collectors
Aggregators, archivers, research crawlers
The long tail of automated collection. Weigh which ones bring value and which just consume bandwidth: with names, not guesses.
Security & Monitoring
Uptime monitors, SSL validators, security services
Legitimate operations traffic, identified and kept separate. Anonymous probes and spoofed visitors are isolated on their own suspicious-traffic report.
New agent in your traffic that nobody has named yet? It lands in a review queue, gets identified, and your history is relabeled automatically.
Verified agents, not claimed names
Anyone can put "GPTBot" in a User-Agent header. Scrapers and scanners do it every day to borrow a reputation. Counting claims as visits inflates your numbers with fakes.
Layer 1
Official IP ranges
Visits claiming OpenAI, Anthropic, Google, Perplexity and others are checked against the IP ranges those operators publish.
Layer 2
Reverse DNS
Operators without published ranges are verified through forward-confirmed reverse DNS on the visiting IP.
The result
Spoofed ≠ counted
A visit that claims ClaudeBot but fails verification is labeled spoofed and kept out of your agent numbers: reported separately, never mixed in.
And crucially, there are three outcomes, not two: verified, spoofed, or unverifiable. An agent whose operator publishes no ranges is marked unverifiable, never accused. Honest data beats confident guesses.
Counting agents is the start. Answers are the point.
A list of bot visits is trivia. Agent analytics earns its place when it answers the questions you would act on.
"Which of my content do AI systems rely on?"
Per-page agent breakdown with colored bot tags: see exactly which articles ChatGPT, Claude, and Perplexity keep coming back to, and which they ignore.
"Do those agent visits send me humans?"
AI click attribution connects the loop: when a person clicks your link inside ChatGPT or Perplexity, BotSights attributes that visitor to the right platform. Agents in, customers out.
"Is my new content being picked up?"
Crawl trends per agent and per page show whether fresh content gets fetched within hours, or sits unread. Catch indexing problems before rankings move.
"What is wasted or hostile?"
Crawl errors show wasted agent visits on broken pages. Suspicious traffic gets its own report with per-network block advice, including copy-paste firewall rules.
Agent analytics vs. what you have today
| Capability | Google Analytics | Server logs | BotSights |
|---|---|---|---|
| Sees server-side agent visits (bots never run JS) | ❌ | ✅ | ✅ |
| Names and categorizes 1,900+ agents | ❌ | ❌ | ✅ |
| Verifies claimed identities (IP + reverse DNS) | ❌ | ❌ | ✅ |
| Separates spoofed from real agents | ❌ | ❌ | ✅ |
| Attributes human clicks from AI platforms | ❌ | ❌ | ✅ |
| Detects social shares via preview bots | ❌ | ❌ | ✅ |
| Dashboard, trends, and alerts without engineering | ✅ | ❌ | ✅ |
Not sure agents can even reach your site? Run the free AI visibility checker first: it tests robots.txt and server-level blocking for 13 AI crawlers in seconds.
The agents you will meet first
Every agent links to its profile: identification, verification, and robots.txt guidance.
Frequently asked questions about agent analytics
What is agent analytics? +
Agent analytics is the measurement of your automated visitors: the search crawlers, AI assistants, social preview bots, SEO tools, and scrapers that read websites on behalf of platforms and their users. Where web analytics counts human sessions, agent analytics covers the other half of your traffic: which agents visit your site, which pages do they read, how often, and on whose behalf? That ranges from Googlebot indexing your pages to ChatGPT fetching an article for a user, to WhatsApp previewing a link someone just shared.
Why can’t Google Analytics see agents? +
Google Analytics only records visitors that execute JavaScript in a browser. Almost all agents fetch pages server-side without running scripts, so they are invisible to JavaScript-based analytics. Agent analytics works from the request itself, which every agent must make.
How does BotSights know an agent is really who it claims to be? +
Every claimed identity is verified: visits from bots like Googlebot, GPTBot, and ClaudeBot are checked against the operators’ officially published IP ranges, and operators without published ranges are verified through reverse DNS. Visits that claim a known agent but fail verification are labeled as spoofed instead of silently counted, so your agent numbers reflect real agents.
Is this the same as measuring my own AI agent’s performance? +
No. Measuring the performance of an AI agent or chatbot you built yourself (success rates, token costs, user intent) is a separate discipline, usually called agent observability. Agent analytics as covered here looks outward: it measures the AI agents that visit your website, such as ChatGPT, Claude, and Perplexity crawlers reading your content on behalf of their users.
Is agent analytics the same as bot detection? +
No. Bot detection is a security discipline: it tries to block unwanted automation. Agent analytics is a marketing and content discipline: it measures which AI systems consume your content, because that is where a growing share of your audience now forms its answers. BotSights does both sides justice: verified agents are measured and named, suspicious automation is isolated on its own report.
What is the difference between agent analytics and AI visibility? +
They are two views of the same shift. Agent analytics starts from the visitor: which agents come to your site and what do they read. AI visibility starts from the outcome: how often AI platforms cite your content and how many humans click through. BotSights measures both in one dashboard.
Free tier available. Setup in 5 minutes.