A long time ago I talked about how I approached rapid prototyping during the development of Belle of the Ball. It's been about two years and a dozen games since then, so I've revised that process a bit. Last year taught me how much going at full-throttle design can lead you down the wrong path. These digressions happen in every design process, but I sometimes went further down those wrong paths than if I had stopped and considered my current trajectory.
For A La Kart, I'm trying something I'll call protosprinting.
Playtest Day: GO GO GO!
On playtest day, I'll revisit my notes from my last playtests. I'll anticipate problems, hopefully optimize solutions, and rapidly iterate new prototypes with minimum necessary graphics and layout. When the playtest comes around, I take feedback, write notes, and test on-the-fly handwritten edits when feasible.
Every Other Day: Rest
After that burst of output, I'll give myself 1.5 times as many days away from the project as I spent playtesting. For example, I spend Monday and Tuesday urgently playtesting, then I won't revisit the game until the following weekend. This also keeps me from designing in a vacuum, without immediate actual play feedback.
So far this kept my playtest days very focused on the immediate tasks at hand, allowing me to refine core systems before digressing into minutia. Not sure if this works for everyone, but I thought I'd share my current process.