Let me explain, simply, what Revit is. A parametric propagation engine that maintains legal-document levels of consistency for any conceivable geometry known to mankind. Its core value (to me at least) is that over the course of a 3-year job, it will consistently take the source geometry and output legal documents — even if the software is updated and upgraded in the process. Revit is able to update its billions of lines of code several times over a project life cycle and output the same documentation every time for a CD set. That level of stability is very hard to achieve, especially given the capability of the software.
To where we sit, BIM data isn't really beneficial to planning. Things like areas and ratios — when room data is more important than having accurate elevations and sections. Schematic design does not demand full BIM, and perhaps even design development. Furthermore, the benefits of rich BIM data are inevitably lost when PDFs are created, though in recent years machine learning models can recreate a BIM model from a PDF.
It feels like because BIM is here to stay we are using all means necessary to mitigate the inherent friction with a massive parametric propagation engine. For TestFit, that meant keeping our own geometry and user experience separate so that speed could be fast (thousands of site plans per second) and interactions real-time. Since we drop off our LOD around Concept / SDs, we have a Revit plugin for our architecture customers to carry forward from TestFit into full BIM authoring.