"....It depends how you count "steps"."

Indeed.

I suppose something like "the number of distinct voltage changes per
cycle" might be a good first approximation of something to call steps
and to count.

In your example I would count something like "3" or maybe "2" or "4" -
I always have trouble with boundary conditions...

In any case, it seems that the device you had was effective.
And the only thing I could imagine as having fewer steps would be a
similar device that didn't have the pause at zero...

And yet it was effective -
I wonder if it would have worked with the light dimmer mentioned
above...