Speculation, as someone who grew up with some of those games:
We could use Heretic/Hexen as the most direct contemporary example of a "Doom++ engine": it had keyboard look up/down, inventory, and the ability to fly up/down. Wiki reckons Heretic sold 500k units and Doom 2 sold 1.2M. Hard to say in hindsight if that was caused by less enthusiasm for a fantasy aesthetic or the decline in enthusiasm for Doom-level tech.
We can also look at how far BUILD-style engines were able to push environments in 2.5d engines of the day. Duke3D had sloping floors and wildly changing terrain that was revolutionary for the time. (Remember the second level in the shareware version, where you basically blew up a building halfway through the level?) Other contemporary BUILD-style titles like Eradicator were able to add crude room-over-room by declaring floating platforms and other special entities without having to build them into truly 3D world geometry.
A third point on the triangle might be Rise of the Triad, which pushed the Wolf3D engine well beyond anything a normal person would expect. Apogee were able to add keyboard look up/down and flight up/down, more verticality in levels by adding floating discs and floors (again outside of the standard level geometry).
And to stretch the triangle metaphor to a tetrahedron, there's also a lot stuff from the Doom source port universe that might have been invented if Carmack & Co. had to push the Doom engine as far as it could go. When you can add new things and linedef/sector actions, that allows a lot of flexibility when you don't have to build the engine from scratch.
All that makes me think that 1994 iD could probably have had a commercially successful game on a Doom++, especially if Carmack kept learning all the portability and networking stuff that made Quakeworld work during the shift to Windows 95. What's much less clear to me is whether they'd be able to catch and/or drive the wave of 3D accelerator development, and whether Epic MegaGames would have just leapfrogged them completely with Unreal. I think the Quake-on-Doom++ timeline might end with Unreal becoming the engine more people built on through the 2000s (instead of all the Q3/idTech3-based games). The talent retention could mean that iD stays competitive on worldbuilding and feel (the Doom Bible shows they had some of those ideas, but the tech wasn't ready for them), and Carmack+Abrash+... could probably catch up with Sweeney+.... We probably don't get American McGee's Alice in that timeline though.
EDIT: The other thing to consider is that Descent was also bringing 6DOF (degrees of freedom) to PC gaming. While it was a portal-based engine like BUILD, it had 3D-modelled polygonal enemies etc. It was a beloved but niche title because it was so challenging to play, but it shows that there's a bunch of technical innovation around at the time, and maybe it gets a larger market share and the technical innovation happens with or without iD?