Nobody will buy your gradually evolving system until several years have passed and it works for some actual customer
The parent was assuming people were already buying Oracle as a database (for example), and continuing buying Oracle as a database, and gradually getting CRM features coming with it.
So people will always be buying the gradually evolving system, and it will always/increasingly usefully be working for some actual customer.
But as a CRM system it has no value to the end customer unless it has a certain minimum level of functionality.
"There's a minimum crew requirement"
"What's the minimum crew requirement?"
"Oh, er, one, I suppose".
It has value as long as it has any value at all, there isn't a certain minimum level of functionality unless you assume either 'absent' or 'broken' is involved. By the time a single shipping useful feature rolls into a production release, it has value to anyone who wants that function. Or, potential value, at least. Like a proto-eye that has value as soon as it's a light/shadow sensitive cell.
It wouldn't make sense to release one basic function as a CRM product - which is why the parent is talking about 'evolving' an existing product up to a higher stack layer, not trying to release a new product at a higher stack layer all at once.
So although you can add features gradually and evolve your CRM offering you will not have a single customer until you have actually added enough features that make it actually usable
Following the model where Oracle tries to build and launch a CRM product all at once, or Google try to build and launch a social network all at once - which is completely the opposite of the gradual improvement up to a higher level which the parent was suggesting. You're arguing "you can't gradually improve from where you are because that isn't releasing a finished product in one go", no it isn't, that's the point.
This sounds a familiar argument - roughly what you're saying is that an eye cannot evolve, it has to be intelligently designed? :|