Visitors who came from an AI answer
Referrals from ChatGPT, Perplexity, Gemini, Claude, Copilot, DeepSeek, Grok, Meta AI and more get their own report, so AI traffic stops hiding inside "direct" and generic referrers.
Visibility Tracker
Website analytics built for AI search. Pageviews, visitors and referrers like any analytics tool, plus the two numbers the others leave out: people arriving from ChatGPT, Perplexity, Gemini and Claude, and the AI crawlers reading your pages.
Launching soon. Join the waitlist and we'll email you when sign-ups open.
example.com · last 30 days
Sample dataVisitors from AI assistants
AI crawler visits
What it tracks
Traditional analytics lumps AI assistants in with every other referrer and never sees crawlers at all. Tracker reports both on their own.
Referrals from ChatGPT, Perplexity, Gemini, Claude, Copilot, DeepSeek, Grok, Meta AI and more get their own report, so AI traffic stops hiding inside "direct" and generic referrers.
GPTBot, OAI-SearchBot, ChatGPT-User, ClaudeBot, Claude-User, PerplexityBot, Googlebot, Bingbot, Applebot, Amazonbot, Meta, CCBot, Bytespider and others: which pages they fetch, how often, and the status they got.
A browser script cannot see a bot that never executes it. A few lines of server code report crawler hits directly, so the numbers come from your server, not a guess.
One lightweight script covers the everyday numbers, so you do not need a second tool next to it.
Including single-page apps: route changes count as pageviews without extra setup.
Which pages people land on and read, per domain.
Where visitors came from, with UTM source, medium and campaign.
Desktop, mobile or tablet, and the browser and operating system behind each visit.
One call records a signup, a download or a checkout, with properties: bombus('event', 'signup', { plan: 'pro' }).
Automated browsers and bots are kept out of your visitor numbers. AI crawlers are counted separately instead.
How it works
Create a workspace and add the site you want to measure. Each domain gets its own site key.
Add it to your site’s <head>. Pageviews, referrers and AI referrals start recording straight away.
<script defer src="https://bombus.visibilitykit.app/bombus.js" data-site="bk_…"></script>
Copy the ready-made snippet for Next.js middleware, a Cloudflare Worker or Express. It skips human traffic and static files and only reports recognised crawlers.
Uses a separate server key, shown once and never put in the browser.
Privacy and consent
A visitor cookie and a 30-minute session cookie, set on your own domain. No third-party cookies and no cross-site profiles.
Add data-consent="required" and the script sets no cookie and sends nothing until your banner calls bombus('consent', true).
Pages are stored by path. IP addresses are not stored.
Consent mode
<script defer src="https://bombus.visibilitykit.app/bombus.js"
data-site="bk_…" data-consent="required"></script>
<!-- when the visitor accepts -->
<script>bombus('consent', true)</script>Call bombus('consent', false) to switch tracking off again.
Pricing
There will be a free tier for individuals, plus plans for growing sites and agencies. Join the waitlist and you will hear about pricing before anyone else.
Tracker reports AI assistants as their own traffic source instead of mixing them into referrals, and it records AI crawler visits, which browser-based analytics cannot see because crawlers do not run JavaScript. It is also a single small script with first-party cookies and an optional consent mode.
ChatGPT, Perplexity, Gemini, Claude, Copilot, DeepSeek, Grok, Meta AI, Mistral, You.com, Poe and Phind, detected from the referrer and from utm_source tags such as utm_source=chatgpt.com.
OpenAI (OAI-SearchBot, GPTBot, ChatGPT-User), Anthropic (ClaudeBot, Claude-SearchBot, Claude-User), Perplexity (PerplexityBot, Perplexity-User), Google (Googlebot, Google-CloudVertexBot), Bingbot, Applebot, Amazonbot, Meta, DuckAssistBot, MistralAI-User, Bytespider and CCBot. Crawlers are identified by their user agent.
Only for crawler visits. Pageviews, visitors, referrers, AI referrals and custom events all come from the script tag. Crawler tracking needs the server snippet because bots fetch your HTML without running scripts.
It is about 2 KB, loads with defer, and sends data with the browser’s beacon API, so it does not block rendering or navigation.
Yes. Client-side route changes are counted as pageviews automatically.
It uses first-party cookies only and does not store IP addresses. For sites that need consent before any cookie is set, consent mode sends nothing until the visitor agrees. Whether you need a banner depends on your setup; we do not give legal advice.
Pricing has not been announced yet. There will be a free tier for individuals. Waitlist members hear first.
Intelligence measures how AI assistants answer questions about your brand. Tracker measures what happens on your own site: the visitors those answers send and the crawlers that read your pages. They are separate products.
Visitors from AI assistants and the crawlers behind them, next to the analytics you already use. Join the waitlist for early access.