On broad downvoting patterns
There are two basic reasons why I downvote questions (and I hope that these are the same for others as well).
The first is to elicit improvement. I do this about $40$% of the time, and generally tend to leave and follow up on comments on such posts.
The second is to reduce the visibility of what I consider relatively unsalvageable questions. These are questions that, for one reason or another, cannot be brought to site standards easily. For example, they could be objectively off-topic (asking for personal advice), or be closable for more than one reason e.g. being a duplicate and lacking context.
When I close for the second reason, my priority is towards the visitors to the unsalvageable post, and not towards the OP of that post. I would like people to answer the next better phrased question and give that their best, rather than attempt to repair what I see as a major flaw in that question's design, something that would take time and effort on both sides to repair and is too unlikely to occur on a consistent basis.
On how I responded to the two categories of downvotes
I would recommend following down and close votes on questions of the first kind. In fact, if one holds a vested interest in such a question , then one would follow it for that reason alone and would want to see it improved and act accordingly. I think a lot of users already do this, and I applaud that. I've recently found myself doing it on a small scale as well, but thanks to my good choice, they improved fairly quickly and I removed them from my list within an hour or so of following them.
However, there do exist posts of the second kind, and I do not see a strong reason to follow these. Some posts transcend into discussions, where the intention is to not really improve the question but instead sort it out in the comments. I don't need to follow that conversation if it isn't about improvement.
So , with that in mind, I would say the following :
If you think a post is unsalvageable, then you should downvote / vote to close it and move on.
If you think a post can be improved, then I would recommend the entire protocol i.e. downvoting with comment, and then following the improvement to completion. This is ideal for situations where one can converse with the OP.
For me, the split between salvageable-unsalvageable that I visit is like $40$ - $60$. Based on one's definition of salvageability, it is difficult to say what those percentages are like for others.
My experience doing this some time back
I did this profiling religiously for $10$ days in $2019$. In detail : I followed every post that I down or close voted in this period, and followed up every such discussion that I could. While that's a small number of days, I still went through about $200$ posts. Some observations :
I had an about $15$% success rate of reversal of downvote or close vote/reopen vote overall. That was doing this blindly, of course. With a better selection criteria, I recently reversed $8$ of $21$ downvotes on various questions that I'd followed up by commenting.
I got plenty of useless messages not addressed to me but just conversations between various users. I can't make out what's addressed to me in the middle of all that.
The followed posts swelled up and I had to spend a lot of time disposing them after some time. Seeing that build up alone made me quit this practice slowly.
After that experience, I would recommend this practice on a selective basis. I think it works wonders when we can figure out (perhaps, with experience) what actually is salvageable and what isn't. Having said that, enforcement or making a mandate of it is impossible unless we are able to come to a consensus on what is unsalvageable and what isn't : another Herculean task for another day.