I really don't think that OS/2 success or failure lay in the marketing (though I certainly thought so at the time). OS/2 had one big problem: IBM's perceived commitment to the product.(1) Heck, IBM wouldn't even ship OS/2 on their own PCs until near the very end. Which brings up another failing: IBM did not control the OEM channels like Microsoft did. Most people will get their first OS as an OEM version. Once they have learned it, it is harder to convince them to switch. Today, we have a situation where it is unlikely that people feel as though they could install the OS. Also consider that Windows 3.1 was very successful, even though it received very little visible marketing.

(1) IBM's perceived lack of commitment is ironic. They supported OS/2 Warp 4 for about a decade after it's release, even though it was a dead-end product from day one. Microsoft has a 10 year support cycle for a thriving OS.