As others have said or implied, I really doubt that average browsing time is a good analog for average attention span.

As others have alluded to, it doesn't distinguish between looking at a website to see if it's what your looking for, realizing it's not, and then leaving; as opposed to finding that it is what your looking for. Even if it did, I really doubt that the amount someone spends on a single website reflects their attention span.

And on the source, the use of footnotes is kind of weird. For some claims, there's a footnote with citation, but for a good number of claims there's nothing. For example, it mentions that "Microsoft found..." and lists of a bunch of numbers, but no citation.