While it may not catch everything, another way I've picked up on various tracking services is simply running pihole and customizing the lists based on permitted dns lookups. It's interesting and troubling just how much of my internet browsing is blocked at the DNS level on a day to day basis with no perceivable impact on what's being viewed. I do wonder how many services might start trying to hardcode IP addresses to get around such things in the future. I'll have to try mitmproxy and see what I might be missing.

Yup I do this too and I agree with your observations, about one third of my home network's DNS request are rejected with nearly no impact on my internet browsing. I've also observed the same thing when browsing the web with uBlock Origin set to block third party scripts by default. On most sites whitelisting ~20% of scripts is enough to make the site functional, the other ~80% apparently being redundant for the purpose of the site's core functionality.

Am I understanding it correctly that the app on your mobile is directly sending the data to marketing companies? They don't even collect it and hand it to marketing companies through a separate channel?

No they dont't usually collect the data and forward it to third parties. With most stuff like this apps will embed a binary SDK which is initialized when the app starts. Because the SDK runs in the app's process it has all the same permission and data access that the app itself has. The SDK communicates with the relevant APIs directly. On the web similar models are often used with third party scripts loaded in the main page context.

For example here's Braze's developer integration guide https://www.braze.com/docs/developer_guide/platform_integrat...