I'm not sure if you're being serious or ironic given the recent press attention about Google Search being crap to ensure that Google Ads makes more revenue.
So, yes, they definitely can game it by focusing on short term unsustainable wins and ignoring all medium and long term negative impacts. Push the AB testing games to their limits, like running the same text 5 times with minor "improvements", so that the statistical metrics no longer apply. Dark patterns that cause long term user value to drop but drives some short term revenue. And so on. Then you get the more fun ones like stealing work from other departments without giving them credit. They eat the R&D cost and you get the revenue win. A good middle manager will even spin it so it looks like their team did the R&D as well.
That said, it all works fairly well when you've got a monopoly in the area (ie: search and ads for Google) but in those cases most everything works fairly well.
I generally agree with your original comment, and I think a lot about Schmidt's quote when operating in a startup context. You're right that it's ironic in the Google context. Search should absolutely go back to being measured on retention and they should disentangle from the revenue team. And I agree that the real death of Google was in the statistical game: 95% confidence in a microscopic metric improvement still lets through serious quality degradation 5% of the time. This gets to another important point: OKRs should be _big_ or they're not worth doing.
I think revenue or any other KPIs makes sense when it's easier to drive the KPI the "right" way versus the "wrong" way. As a business grows the low hanging fruit are used up and things start to shift to the "wrong" way being significantly easier. Moreover if the culture is about driving a KPI then, like any culture, it's very hard to shift it and no one wants to eat the political cost of trying to do so. I think executives often forget that people are clever and will game any metric if it's possible. I was at a company where there was a silent agreement across the EMs to over-level engineers. The pay was below competitors at the same level and set by HR but the promotion committees were not run by HR. Every new EM that joined was horrified at first at the competencies of their engineers, but then either learned to play the game or eventually got pushed out by their ever growing enemies.
So, yes, they definitely can game it by focusing on short term unsustainable wins and ignoring all medium and long term negative impacts. Push the AB testing games to their limits, like running the same text 5 times with minor "improvements", so that the statistical metrics no longer apply. Dark patterns that cause long term user value to drop but drives some short term revenue. And so on. Then you get the more fun ones like stealing work from other departments without giving them credit. They eat the R&D cost and you get the revenue win. A good middle manager will even spin it so it looks like their team did the R&D as well.
That said, it all works fairly well when you've got a monopoly in the area (ie: search and ads for Google) but in those cases most everything works fairly well.