I expected these numbers to be higher. However, an even more interesting metric is the 88% block in Firefox.

Firefox may not have a great market share, but based on these numbers it's market share may very well be eight times higher than your analytics report. This can change the argument of "it's only 3% of our users so we don't need to test on FF" to "it's a quarter of our user base, we should at least test it", depending on your target audience. I've seen tons of people claim general Firefox usage is negligible based on public data from websites such as statcounter, but these metrics prove that those statistics are unreliable and should not be used.

The best you can do is use server side UA inspection, though you can't really distinguish bots from real users that way.