One of the notable things about FAANG processes that I've observed from my friends there is that roles and processes are mechanized[0]. Individuals are placed like precision robotics in a bigger machine. This kind of structure means that you have a defined process for promotion or pay raises and you know what your role is. In fact, one might even posit that the ability for these large organizations to create a machine to extract surplus from labour in a systematic fashion is the reason for their success.

For most people, this is wonderful. Knowing what you will be valued for is very useful. It says "do the things that are useful to us" and "stop doing the things that are useless to us" and tells you "these are the things that are useful and those are the things that are not useful". At their scale, rare errors in the process will inevitably show up, but smaller companies often have these errors at higher rates. All that to say, success often comes from identifying what is useful to the organization and what is not, and then what is useful and what is not to the person who has control over one's role in the organization.

In mechanized organizations, this should be easier. In unmechanized organizations, one's skill at this will dominate one's technical skill at determining success. But it's just a skill, and if you cannot find a way to train it, the easiest workaround is to ask the person making the decision: "if I wanted it in 3 months, what would I have to do?".

You may get an answer that was untrue 3 months later, but you just shrank your timeline in a way that is much more meaningful, and perhaps more likely is that you'll either get an unrealistic timeline (which is useful signal), or you will hit it and get what you wanted (which is also desirable).

0: A classic example of this is that no one can "get you into Google/Facebook/whatever". This reveals the other non-obvious purpose of their interview process besides quality-control of hires: quality-control and rules compliance on interviewers.