Most of the times, it is recommended that you test with only a few number of testers, because when you are testing for the very first time, a lot of information you get from watching the first tester is brand new, and when you are repeating the test with the second tester, a lot that you saw before with the first tester will be repeating, and so on for the next testers. The less users you have, the more rounds of tests you’ll have, because you will have more time for that.
Still, there are a few situations that you may need many testers, for example, when you want to make a quantitative study for the usability of your program, the more testers you have, the more precise your results are.
Usually, for those kinds of tests made only to improve usability, the number of testers is 5. But why 5? According to this article, each user can find an average of 31% of usability issues, and each time you test you’ll be finding new issues, but from the 5th user, you no longer have significant results from other tests. Of course that if you test with more users, you will find more issues, but as you add more users, you learn less and less because you’ll keep seeing the same things again and again.
Using the function proposed in the reference article:
where L is the amount of issues (assumed as 31%) and N is the number of testers, we’ll have
Note that with 5 testers, we’ll only have 85% of the usability issues found, but as we increase the number of testers after 5, the usability issues found approach very slowly to 100%. So, five testers is said a good number to have enough information to improve usability and test again and again in another iterations.