Continuing Michael Bernstein - CS 197 - Vectoring, Michael Bernstein suggests that once you’ve figured out that most important question to answer, you must next make very rapid progress. You’re aiming for one-week cycles, ideally with multiple iterations in each cycle.
If velocity = distance / time, the typical problem is that the time component is too large. You’re addressing questions that are too broad, or you’re being too perfectionist. Get scrappier.
Velocity can be a freeing way to think about how your project is going, because unlike progress, it’s under your control. If you’re stuck in a “swamp”, what you need to do is to rapidly try lots of stuff. Then you might accumulate enough insight that you can actually make some progress. But you can’t directly control progress. Another way to put this is that velocity is an input, whereas progress is an output.