In this question it is asked whether subgroups of a finite, 2-generated group are also 2-generated. The answer is no, with a nice counterexample. However, when the group is Abelian, this is true. My question, albeit a bit open ended, is what are some other nice conditions under which this is true? I have a particular example where it turns out to be true, and I'm trying to understand why. Two properties that seem to be relevant in this example is that the orders of the two generators of the large group are coprime, and that although the group is non-abelian, every element is of the form
$$a^ib^j$$
Where $G=\langle a,b\rangle$. Are either of these conditions sufficient to guarantee the claim? If not, what other conditions can be added to guarantee it?