The number to compare to would be the fraction of "readers" who actually read (most of) the article. I have a hunch it's also a small fraction.