I also realised a while ago that most emails are just transactional now (password resets, how well did we do the thing you paid us to do?), so I do similar to you without the sender input and it more or less works. Breaks down like this:
- Nobody gets into my inbox unless whitelisted address. Instead of fighting spam and blacklisting, I know my inbox is always from someone I want to hear from.
- Every other email goes through spam assassin (which is getting worse) and then into a folder called Transactional (and if it makes a mistake, I add to the whitelist)
- I have another folder SpamAssassin where I move spam, and it gets pushed back through sa-learn
- Finally, I have a cronjob that goes through the email in Transaction and looks for subjects, senders, sending domains or to addresses that have a) been received at least 5 times and b) are only in the spam folder.
So kind of the same idea as you. But I think I'd feel really pretentious using this system and assume that quite a lot of people just rolled their eyes and ignored the rejection.
this sounds really similar to how the tool works and processes emails, too.
Would you feel this would be more helpful if the tool just silently manages the emails and does not send any replies to the original sender then?