Librarianon

I'd be collecting big data shit on most popular builds, most popular legendary weapons, most chosen skills, most popular implants, car purchasing habits, when do most people drop off the game, how many players use trunk stash, does anyone actually craft non unique weapons, etc..

When I worked on Thief (2013 (yes, the bad one)), publisher Square Enix collected all sorts of pointless bullshit stats in the name of Big Data. Things like how often players fired each arrow type, how much loot they collected, and how many guards they bonked. Note that this was done for a single-player game, so none of this data was gathered in the name of long-term balancing. Similarly, another AAA game I worked on that came two years ago phoned home about the footsteps of all players at every level of the game. The problem with this gluttonous data-gathering obsession is that it's very hard to formulate a coherent narrative from your data when all you have is gigabytes of noise.

Free-to-play game companies do this much better; they'll start with a problem statement, like players aren't opening enough lootboxes. Then they'll formulate a theory, like, what if we painted the existing boxes blue? Only then do they gather data to prove or disprove the theory. For example, they'll do an A/B test with 1000 players, and they see that coloring the lootboxes blue increased conversion rates by 0.5%. Obviously, there are limits to this approach, and F2P companies regularly get stuck in the mud researching stupid shit like the "optimal" color for a lootbox. But comparatively, the AAA game companies I've worked at are stuck in a mindset that any data is better than no data, which is just objectively false. Shit data is much worse than no data.

And no offense to the OP because game telemetry is a fascinating topic, but the most important stat in that proposed list is "when do most people drop off the game." You can track that very easily, with achievements that unlock after every chapter, for example. But even then, that data only matters to the business if you're making a sequel or if you have a way to monetize players after the initial €60+ sticker price.