Ishant Juyal’s approach
So the guardrail metrics that I would choose here would depend on the risk that this change would lead to. If we increase ad impressions per user, the risks are that people will skip ads faster, do more rage clicks. They might even stop videos. They might stop coming back if the UX gets too bad, and in general declining user engagement. So I'll maybe, as we are tying it to user engagement, and for YouTube the right user engagement metric for me would be the amount spent per user per day. And that is the guardrail metric that we need to account for, that if this metric moves below a certain value or changes negatively by a certain percentage, it's bad for us. So the first guardrail metric would be the average view time per user per day. And second is the video abandonment rate. But that doesn't really make sense. So maybe something like, yeah, like abandonment of video is again leading to loss of user engagement. But I am still confused a bit in another one, which is like ad skip rate. So these are my choices.