One of the things that is missing from the comments here that say "Office Substitute X does 99% of what users need" is that in most companies, there is already a way of doing a task.

In many complex tasks, Excel is a front end to a complex business process. That business process is kicked off by some VBA code that talks to some COM object that talks to the back-end infrastructure. For example, in Finance, there are scenarios where that COM object kicks off a job on a Linux HPC cluster to compute, for example, the risk of a position that a trader is interested in. While "Office Substitute X" could potentially do the same thing, the trader has absolutely zero interest in using anything other than Excel, and he/she is the rainmaker.

So there is very little cannibalization of the existing desktop Office business for Enterprise customers. What this seems to be doing is taking people's familiarity with Office at work, and making it cheaper/easier for them to continue to experience Office in their personal lives.