I actually decided to self host analytics and generated a simple drop in google analytics replacement. People overthink these things. It's a very straightforward analytics API. And if you ingest the data in a good database or metrics engine (I used Elasticsearch), you can query it quite easily.

In my case, my motivation was that I want to use LLMs to query the data with agents. This whole thing was surprisingly easy to setup and a positive thing is that you don't have a scary extra data controller doing shady things with the data.