Open Analytics

Hey Product Hunt 👋

We’re the team behind Open Analytics.

We built it because we kept hitting the same wall: every analytics tool made us choose between knowing our numbers and respecting the people behind them.

Google Analytics gave us the numbers, but often came with complexity and consent banners. Many privacy-first tools removed the banner, but stopped at pageview counts. We wanted the best of both worlds.

Open Analytics is cookieless by design. Visitor identity is represented by a salted hash that rotates at midnight, raw IP addresses are never stored, and no cross-site profile is created. The result is useful analytics without persistent tracking.

Here’s what we’re most proud of:

• Privacy that’s enforced, not just promised. Global Privacy Control is honored by the collector server-side. Even if the snippet is configured to ignore it, those requests are dropped before they’re counted.

• Revenue attribution where the visit happened. Connect Stripe and each payment appears alongside the visit that generated it, showing which pages, channels, and referrers drive revenue.

• Analytics you can ask questions about. Use AI Chat inside the dashboard or connect through our MCP server to explore your traffic in plain English.

• Custom events without custom JavaScript. Add data-oa-event="signup" to a button and Open Analytics tracks it.

Open Analytics is open source under AGPL and fully self-hostable. If you’d rather not run it yourself, our hosted version starts at $9/month. Every plan includes every feature—you pay for volume, never for access.

We’d love your feedback.

Building an open-source, AI-native alternative to Google Analytics is such a great move. Congrats on bringing this to life!

Love that you made analytics both cookieless and MCP-ready, giving me real insights without a consent banner while letting my AI tools actually query the data.

My dashboard told me a subscriber had churned. Turned out to be my own test account, I refunded it myself and the rollup had no way to know those were the same person.

And the counting was all correct. The mistake was one layer up, in what I'd told it counted as a person, so no query would have caught it.

So with the chat window and the MCP server sitting on top, does it show you the query it ran?

@uaghazade The identity hash rotating at midnight is the honest way to do it, but it also means someone who comes back tomorrow looks like a brand new person, right? How do you handle funnels that take a few days, like ad click on Monday and purchase on Thursday? That's the report I always get asked for haha.