Agreed, having been on the inside at Google - they are not good at producing customer impact. The orgs are so large, so diffuse, and upper management so disconnected from on-the-ground execution that it's really all about gaming metrics. Even people who still want to produce customer impact have to play the game of stuffing their feature into a metrics-shaped-box so their projects don't get shut down.

Google is obsessive about engineering quality - and IMO is a good example of why that shouldn't be the core principle of a tech company, even though as engineers we're biased towards it. Google excels at producing a lot of well-designed, scale-hardened, well-tested code... that doesn't do anything useful for the business and is a net drag on the company.

re: customer impact, and if you want to define "real engineering work" as "real engineering work that your users actually use/love", IMO Apple is the way to go in the realm of BigTech. It's the only one of the FAANGs where simply juking metrics isn't enough to get ahead, and you have to deliver customer impact in a reasonably robust way to succeed.

I've been at 3 FAANGs now, and Apple is the only one where a conversation has gone something like "wait, the metrics you're proposing are bad at measuring the actual quality of the user experience and does not account for hard-to-measure factors like (a), (b), or (c)", and that conversation is coming from an executive.