Data I/O, the company I worked at, bought a few copies of CPM/86 and we tried it out. There were a few people who liked it, but most preferred DOS. Nobody could explain to me any reason to prefer CPM/86 - technical, practical, or price.
CPM/86 had a shot. IBM sold both. CPM/86 lost fair and square.
The miscalculation is that IBM PC sales were mainly to first time users who had no history with CP/M, and (as I mentioned) the technical superiority of CPM/86 was not apparent.
But a significant problem with CP/M was that if you swapped a floppy disk in a drive the system wouldn't notice it and things could get really messed up, so you had to remember to always control-c after changing disks. DOS would check, which reduced performance but was much more user friendly.
The CR-LF line endings also trace their way back to DEC.
Hmmm...were you there when Data I/O was planning on acquiring Callan Data Systems? If so, what did you guys hear about the reason the acquisition did not go through?
I was at Callan at the time as a Unix kernel hacker, and we technical people never heard any solid information on why that acquisition did not happen.
Probably more accurate to say that no single person knows the whole story. And those that know parts of the story can't necessarily be trusted to tell the truth, either from forgotten facts, or deliberately self serving commentary.
If you look at all of the stories from everyone, the facts that they all seem to agree on is that IBM went to Microsoft first, then went to Digital Research, and Microsoft came out a big winner. You could eventually buy an IBM PC with CPM, but at a much higher price. The reasoning behind those events are what's up for debate/speculation.