This is a great read about the rise and fall of WordPerfect. http://www.wordplace.com/ap/
The latter had too much pride in their collection of DOS printer drivers, but what really killed them was releasing something that passed the bar of not crashing, but would per a lawyer I worked with regularly drop all your tables and figures to the bottom of the document. Borland was slow to get their stuff working on Windows, was late to bundling an Office equivalent according to a marketing guy from there I once worked with, and had the bad portent of the edifice complex: http://www.krjda.com/Sites/BorlandInfo1.html
To this day, well it's after they started going beyond their start in computer languages, their 808x C compiler for example blew away the competition when it finally launched, we ding Microsoft for their quality problems, but especially in the days Gates ran it one of the "secrets" of their success was writing software that basically worked, didn't crash on you etc. This makes it easy to beat competitors who lose that ability, or aren't as good at it.
Actually, this bet was made because Microsoft told Lotus OS/2 is the future. They developed a very nice spreadsheet on OS/2, but we know how that ended.
While Lotus was focusing on OS/2, M/S created their spreadsheet for Windows in secret and stole the market. Lotus did not have the resources to develop on 2 different platforms.
This is from people I knew who were high up in Lotus. Another reason to never trust M/S now that they "like Linux". :)
My take on it is that Microsoft was at first enthusiastic enough about OS/2 but IBM doomed it, first and foremost through their "the PC AT is the last personal computer you'll have to buy" message to their customers, something they used to be able to do when they controlled a lot more of the business computer market and "no one ever got fired for buying IBM." Forcing OS/2 to run on the 286 was Not Even Wrong, for example in protected mode a 64KiB segment change incurred a hit of many cycles (40?), Intel did not understand how people used the 808x and only fixed that with the 386.
But did your people "high up in Lotus" mention how the mid-life kicker for 1-2-3 using expanded memory (https://en.wikipedia.org/wiki/Expanded_memory) was coded by people against orders who'd left the company by the time it was rediscovered and made into a product?
That the Jazz office suite for the Macintosh was developed using assembly "for efficiency" but without a LAN or other means of close collaboration, every week people would hand in floppy disks for a complete build and thus a great deal of duplication made it very much less efficient? And in general how poorly that software development model worked? I see in Wikipedia "in Guy Kawasaki's book The Macintosh Way, Lotus Jazz was described as being so bad, 'even the people who pirated it returned it'."
I was working in the Boston are in the 1980s and knew a lot of people who'd worked for Lotus or coworkers or friends of them, so I got takes from their very beginnings, which was a handful of people writing in close collaboration 1-2-3 in, yes, assembly language.
That gave them a couple of years when Microsoft applications could do things on standard customer machines that their competitors couldn't do.
That one deliverable was worth more than MR has cost up to this point.
And/or make better use of the extra segment and maybe stack segment, compiler technology there. I'd imagine other compiler hacks would be useful, for example compare to today's bigger systems where the premium is on having as much in the most local caches as possible. Cleverness once you're in them is overwhelmed by the penalty of going to a more distant cache, memory, or taking a TLB miss. "Cache is the new RAM, RAM is the new disk, disk is the new tape." and all that.
It was a viable competitor of Microsoft word in the Ms/DOS era, but did not manage the transition to the windows era.