Isn't that the point? Something like 80% of their revenue is from ads so that's where their focus lies. As cool as a lot of the other products are, they just don't seem to do much for the "business" other than supporting the data for targeted ads. Meaning, anything that does support their main revenue stream is at risk during economic downturns. (I'd be happy to be proven wrong by someone who, unlike me, has real insight)
I’d use GCP as an example of the problem: we buy services but their sales guys showed up like “we’re Google, of course we’re the best” and put approximately as much effort into selling as they do for Gmail. AWS makes a ton of people available, listens to what you need (and actually ships it), and follows up. Guess who gets the sale? Anthos was a cool idea, lots of smart people worked on it, … and none of that matters if you give up after begging them to show up to sell it.
The Google+ and messaging debacles are obvious examples but I think GCP and G-Suite were also seriously held back in a way which an independent subsidiary which had to make a profit on its own would not have been. Their executives just weren’t sweating the way they would have if they knew their shares would go down when they didn’t execute well. Contrast with AWS, whose leadership are running the service in a way which makes it clear they have no plan B involving ongoing subsidies from the parent company and thus aren’t blowing off customer support, features, security, etc.
Chromecast is definitely good but it was years after Apple and Roku, and feeds data to the ad impression system so I’m not sure how much it defined the market rather than very successfully competing in it.
Google’s educational efforts are actually super interesting to consider: they had some initial success but seem to have lost their way. My wife is a teacher so we hear people talking about this: most of the districts around us have switched to Office 365 and almost all of the people who’ve mentioned that have been happy because it’s a better product without so many of the never-fixed warts they hit with G-Suite. That feels eerily similar to what I see with GCP where you really get the impression that incremental improvements aren’t favored by whatever incentives their management sets.
I think part of the distinction is how "success" is being defined. A product may be innovative and market defining, yet still not contribute much to the bottom line. It's still cool to work on, but these companies are still trying to make money at the end of the day. I assume their moonshot ideas still have the explicit long-term goal of being profitable.