That may not have mattered had OS/2 done a good rapid development kit, and also done a good job getting OEMs (there were hundreds at the time, in contrast to single figures today - aside from Dell, HP, IBM (now Lenovo) and Acer, there were Gateway 2000, Micron, Zeos, MidWest Micro, Tagram, and hundreds of other PC vendors.
I think OS/2 could have been a success had it been in the hands of someone other than IBM at that stage - a smaller company whose very success depended on OS/2 - as opposed to IBM, for whom it was just another small piece of a puzzle lost somewhere in the store. Such a company could have capitalized on the anti-Microsoft sentiment amongst ISVs, a number of whom found themselves competing against Microsoft despite making products that helped in the success of Windows. Companies like Borland, WordPerfect, Lotus (before IBM gobbled it), Symantec, and a good number of others. Such an approach could have ended up first in viable tools for OS/2 (from Borland, Symantec, Watcom et al) followed by viable apps, like Lotus SmartSuite, WordPerfect Office Suite and others. Given viable apps at the time, a good number of vendors would have offered the choice of OS/2 to customers, along w/ Windows. In fact, during the long delay in launching Windows 95, OS/2 could well have filled up the vacuum - just like Linux filled up the vacuum while the UNIX wars were going on.
The other great mistake that IBM did was pulling the plug on OS/2 for PPC. By the time it happened, it was obvious that OS/2 for Intel was going nowhere, so a good strategy would have been to port OS/2 to the PPC, and make it available as an alternative to MacOS for Mac clonemakers, when Apple, w/ a newly returned Jobs, pulled the plug on them.