Mainly for those reasons:
1)
I want a mail system that is so-called KGS (Known good Set) i.e. several core components (postfix, clamav, etc) just work well together and don't step on each others toes
2)
fast deployment
3)
repetetive deployment which guarantees to cater fot the exact same result e.g. depoying the same setup for 2 or more customers
4)
something that provides email as well as mailing lists
5)
project needs to be active i.e. forseable and ongoing development
6)
have the full participation of the technical community and hence knowledge base benefits.
7)
It uses mainstream, tried and tested components. For example, there is citadel, a great piece of software but then I want mainstream components that's why I am here ... at iRedMail ...
8)
has to work flawless within OpenVZ VEs (Virtual Environments) http://sunoano.name/ws/public_xhtml/openvz.html It seems iRedMail considers that important which is good. That's why I am here. That's why I am going to stay.
9)
stable i.e. in case of Debian, if it picks from stable repos, that is good ... peace of mind. Also as mentioned, it's good iRedMail is more or less a smart metapackage since, if all goes wrong, the power of the underlying distribution makes it possible to not get in huge troubles because of some possible SLAs (Service Level Agreements) one might have with customers.
10)
as mandatory as is the must-be support for OpenVZ, for us things need to work with Debian. We are on a 100% Debian infrastructure. That is not going to change. Actually I think it's quite the opposite ... the number of Debian installations resp. free software in general is going to see major boosts in the future. Just recently I read an article that China has now become the biggest "User" of free software. Imho we will see an explosion of free software usage within the next two or so decades.