Web analytics that shows your AI traffic
See the visitors ChatGPT, Perplexity and Google send you, and the AI crawlers that read your site.
Open source · No cookies · Free during the beta
- Search
- Direct
- Social and referral
- AI answers
- Google18,240
- Direct9,871
- ChatGPTAI2,412
- GitHub1,985
- PerplexityAI812
- ClaudeAI371
- GPTBotOpenAI81.2k
- ClaudeBotAnthropic44.9k
- OAI-SearchBotOpenAI38k
- PerplexityBotPerplexity19.6k
- ApplebotApple8.9k
- ChatGPTsent a visitor to/docs/install
- GPTBotcrawled/pricing
- Googlesent a visitor to/blog/rate-limits
- ClaudeBotcrawled/docs/api
- Perplexitysent a visitor to/compare/stripe
Your next customers ask an AI first
AI now reads your docs and blog before people do. Most analytics tools drop AI crawlers as bots and lump AI answers into referrals. Micaforge shows both. Why AI traffic matters
See which AI answers send you visitors
ChatGPT, Perplexity, Claude, Gemini and Copilot get their own channel next to Search and Social, with the same goals and conversions.
| Channel | Visitors | Sign-ups | Rate |
|---|---|---|---|
| Search | 19,840 | 412 | 2.1% |
| Direct | 11,920 | 301 | 2.5% |
| Social and referral | 8,642 | 139 | 1.6% |
| AI answers | 3,904 | 187 | 4.8% |
On acme.dev, visitors from AI answers convert at 2.3× the search rate.
Know every AI crawler by name
Each crawler is named by company and purpose and checked against the IP ranges that company publishes. Fakes are flagged, and you see who ignores your robots.txt.
| Crawler | Crawls | Identity | robots.txt |
|---|---|---|---|
| GPTBotOpenAI · Training | 81,200 | Verified | Follows rules |
| ClaudeBotAnthropic · Training | 44,900 | Verified | Follows rules |
| OAI-SearchBotOpenAI · Search | 38,050 | Verified | Follows rules |
| PerplexityBotPerplexity · Search | 19,610 | Verified | Follows rules |
| ApplebotApple · Search | 8,930 | Verified | Follows rules |
| ClaudeBotClaims Anthropic · Unknown | 2,140 | Fake | Fetched blocked paths |
Find the pages AI reads but never credits
For every page, compare how often AI crawled it with how many readers came back from an answer. Then write more of what works.
- /docs/api/authenticationNot credited
- /blog/why-cookielessCited
- /docs/installCited
- /pricingNot credited
Everything you used in Google Analytics, without cookies
One script and one dashboard for the reports you check every day. Compare with Google Analytics
- RealtimeWho is on now
- PagesTop, entry and exit
- SourcesReferrers and UTMs
- LocationsCountry to city
- DevicesBrowser, OS, screen
- Custom eventsWith properties
- GoalsConversions
- FunnelsFind the drop-off
- JourneysPaths between pages
- RetentionWho comes back
- Web VitalsBy page and device
- Reports and alertsEmail, Slack, webhook
- Feature flagsRoll out safely
- Shared dashboardsPublic or private link
- Stats APIAnd CSV export
- ImportsFrom GA, Plausible, PostHog and more
Private by default. Live in two minutes.
- No cookies and no consent banner
- No personal data or IP addresses stored
- Script under 3 KB, never blocks your page
- Your data stays yours: export it any time
- 1
Add your site
Sign in with your email. No password.
- 2
Paste one line
<script defer data-site="1" src="https://api.micaforge.com/mf.js"></script>To count AI crawlers too, add the edge middleware or ship your server log. Set up crawler tracking
- 3
Watch visitors arrive
The first visit shows up in seconds.
Open source. Self-host whenever you want.
Micaforge is AGPL-3.0. The hosted version and the self-hosted build are the same software, with no paid features held back. Run it on one small server with Docker.
Frequently asked questions
Does Micaforge use cookies?
No. Visitors are counted with a salted hash that changes every day. No cookies, no raw IP addresses and no personal data are stored.
Which AI tools and crawlers does it recognize?
Visits sent by ChatGPT, Perplexity, Claude, Gemini and Copilot, and crawlers such as GPTBot, OAI-SearchBot, ChatGPT-User, ClaudeBot, PerplexityBot, Googlebot, Applebot, Bytespider and CCBot. Crawler claims are checked against the IP ranges each company publishes.
How can it see crawlers that do not run JavaScript?
The script counts people. Crawlers are read from your server log or from a few lines of edge middleware that report each fetch to Micaforge.
Is it free?
The hosted version is free during the beta. The software is open source under AGPL-3.0, and you can self-host it for free.
Can I bring my Google Analytics history?
Yes. You can import your history from Google Analytics, Plausible, Umami, PostHog and Simple Analytics.