This is an interesting question.
In general, VO campaigns do well for apps that have a wide distribution of LTV values -- that is, apps in which it's possible for users to spend lots of money and become "whales." Those whales drive up overall campaign ROAS and subsidize free users.
So with a VO campaign, you're setting a Day 1 or Day 7 ROAS target and Facebook is hunting for the highest-value (see this question for background: How does Facebook’s advertising targeting algorithm work?) users to reach those ROAS goals. When targeting is set really broadly, Facebook has a large pool of users to pull from, but ultimately there's some finite number of people that Facebook can reach that will 1) enjoy the app enough to engage with it and 2) will engage enough with the app to want to monetize heavily with it. At some point, Facebook will reach some substantial proportion of those people and CPMs will go up, meaning install prices will go up, and Facebook will have to reach ever-greater monetizing users in order to hit ROAS goals or delivery will slow down.
Compare that with AEO on a purchase event, where you're only optimizing for that one event, not some magnitude of overall spend that needs to meet a ROAS target. So with AEO, Facebook can onboard a bunch of people that all do that event and the campaign is successful; with VO, a campaign is only successful if Facebook onboards enough whales to offset the cost of the non-spenders. There are more users in Facebook's universe that will make one payment than there are users that will make many payments or make large payments.
So in the long run, I tend to think that AEO campaigns that are optimized for purchase events (given that those events have been systematically tested for signal) will outperform VO campaigns at scale on a ROAS basis. I say at scale here because meeting ROAS targets isn't really impressive / helpful if scale deteriorates to a trickle.
In general, I don't think it makes sense for advertisers to use VO campaigns unless the app's appeal is very broad and the app's monetization supports a very long-tail LTV distribution (ie. it's possible to spend thousands or tens of thousands of dollars within the app). I have not seen VO campaigns work well on eg. subscription apps because there's no magnitude to LTV in a time period that the advertiser cares about (usually, 1 year): users either purchase or they don't, and when they purchase, it's within a limited range of prices. VO campaigns are targeting users based on magnitude of spend, and what I've seen happen over time is that install prices get so high that even when Day 1 or 7 ROAS targets are being hit, making up the install cost over 60 or 90 or 120 days is impossible.