I seek to min-max the subjective value retrieved from the service/product vs. the amount of data given to the data vultures, while taking alternatives into account.
Some examples, including the ones from the OP:
Lemmy - I’d probably still use Lemmy even if it vultured some of data, as the perceived usefulness of its service is fairly big. However I’d probably start implementing measures against said data vulturing, including: avoiding to speak about certain personal subjects, switching accounts every 2~3 months, bullshitting a bit in the comments, etc. Just like I did in Reddit.
Facebook, Twitter - they harvest so much data that even if they handled users better I’d probably not use them, plus the benefit that I get from their idiotic userbases is fairly small. I’m still fine however opening some link here and there, I just don’t actively interact with those platforms.
Youtube - I’ve been using it nowadays through piped.video, to minimise the amount of data exposed while still watching my videos. If piped didn’t exist I’d probably watch it straight from Youtube, it’s a privacy nightmare but more acceptable than lack of access to videos.
websearch - I look for things in DuckDuckGo first. If DDG didn’t find it, only then I use Google. No Google account, incognito mode (to clean cookies afterwards), location spoofed to Antarctica (mostly due to result relevance).
all - uBlock Origin, canvas blocker, decentraleyes, location guard are always active. And if a site doesn’t work well with those, odds are that it’s wanting too much of my info to be worth my attention, specially if I can get its services from elsewhere.
So when it comes to privacy I try to keep myself safe/private but without going full zealot about it.