Factors to consider:
In-house customization may not always be supported by the same CAD support personnel; people quit, move away, get fired. Few CAD support personnel are hard-core programmers, and many don't even know the basics of .NET. Most CAD support personnel, however, *do* know at least the basics of LISP. Future supportability is key, and leaving behind programs that cannot be maintained by anyone else is bad form.
For professional 3rd party applications development, there really is no excuse for not using ObjectARX/.NET. Professional software companies employ programmers, not CAD support specialists (which is why some software doesn't work exactly as we would like), and they should have the skills to create efficient and robust code that includes proper source code protection (something which is lacking in LISP and VBA).
Speed is also an issue. I'm still learning .NET, but even if I knew it backwards and forwards I would still do the majority of support in LISP or VBA, simply due to the speed of creation and phenomal depth of resources for its use in AutoCAD. Compare the number of LISP entries here with the number of .NET entries.
Available development tools are also an issue. If it comes down to it, I can develop LISP programs on any computer with Notepad. I don't require the .NET framework, any kind of code compiler, not even AutoCAD. In theory, that could be done with say C#, but its much more difficult without the object browser, .NET references, word-completion, and all the other tools the IDE requires for good .NET programming. Also, because of the simplicity of LISP, debugging is much faster and simpler.
----------------------------------
If you are going to fly by the seat of your pants, expect friction burns.
"I don't know" is the beginning of knowledge, not the end.