We were able to import the design into KiCad and produce boards that worked. The board was a mix of power and signal, 4 layers of 70 um copper. Here are the main issues we ran into:
Needed to update our database library to handle both Altium and KiCad.
Lack of padstacks caused a few small issues with some SMD mounted threaded standoffs.
Lack of outjob-type capabilities means more complicated production of output files that fit our design process.
These were not showstoppers. We have several permanent (we hope) Altium licenses and are still paying maintenance. However, we were in addition paying for a few monthly seats. We have ended that practice since KiCad suffices for those. In addition, we are not adding seats as we grow, since KiCad meets our needs at this time. We use an internal parts database library and have structured the library to work with both Altium and KiCad. This cost us some time and effort, but was worthwhile and now it is a simple matter to add new parts. We do make separate footprints and symbols for KiCad and Altium, but at this point the main work is done and the incremental cost of adding parts is low.
Every year, KiCad improves greatly, though keeping up sometimes is extra trouble. This is unavoidable given the goal of becoming a feature-complete professional-level tool. Altium doesn't change much, and most of the changes seem to be in document management. We don't use these changes as our process is set.
So, short answer: we are not getting rid of existing permanent Altium seats, but new seats are going to KiCad.
John