Unless the quality of life is reduced because they have a false belief that it would increase, and I can prove to a sufficient degree that that belief is false, I'd say "it's pretty much never wrong". Is it really wrong to get paid for constructing a church if I think organized religion makes people worse off? And I certainly wouldn't want someone applying that principle to me, because I know how bad other people are determining what reduces the quality of my life. It's this logic which leads to large soda bans.

I usually approach this sort of question by modeling people as approximations to ideal agents who reliably act in such a way as to actually optimize the world for their own values.

If I consider a hypothetical person who very closely approximates that ideal, I'd say it's generally+ not at all wrong to enable them to significantly reduce the quality of their own life... they will either do so, or not, depending on their own values.

If I consider the other extreme, a hypothetical person who reliably acts in such a way as to optimize the world for the opposite of their own values, I'd say it's generally+ wrong to enable them to make any effective choices at all.

Either way, their happiness is largely irrelevant to me except insofar as it's subsumed in their values, and whether I benefit from their actions is irrelevant.

That said: obviously I'm more inclined to motivated cognition when I benefit, and therefore need to be a lot more scrupulous about whether my thinking has gone completely off the rails.

There's something to be said for the rule of thumb that if a line of reasoning tells me it's OK for me to act in ways that predictably lower the quality of other people's lives and benefit mine, I should reject that line of reasoning as flawed... not because that's necessarily the case, but because human minds being what they are that's the way to bet.

+ There are exceptions in cases where I think their values are themselves wrong, but I think that's a different conversation.