There are a number of problems with this - it is over all service plans and includes home links to the end user computer (usually wifi), but Netflix believes this more or less averages out and the relative ranks are meaningful. I've seen several more detailed studies and the relative ranking isn't too much of a surprise.
Of course the rub is you can rarely get more than two providers in any region and sometimes it is one or even none ... But if you are thinking of switching ISPs it may make sense to see if there are big differences between those you have available.
This sort of testing should be encouraged and at a level where direct numerical realistic performance characteristics are measured. It would be a useful tool for competition. And, of course, it makes sense to do this internationally.
A correspondent notes that Netflix requires at least 2 megabits per second for standard definition movies to play without long pauses for buffering - so if your ISP is not better than that, the quality of your experience will suffer. It is embarrassing to see so many numbers this low.
I'm sure this will be met with comments suggesting Netflix isn't measuring properly, but let's have some transparency and have proper regular monthly measurements for all of the providers in all of their regions.
Perhaps this is the only way we get service improvements.