Regarding the differential equation $$ y'' + p(z)y' + q(z)y = 0,\quad z\in\mathbb{C}, $$ we can find solutions of the form $$ \sum_{n=0}^\infty c_n (z-z_0)^n, \quad c_n\in\mathbb{C}, $$ given that $p(z)$ and $q(z)$ are analytic in $z=z_0$. Here $|z-z_0|<R_1$ for some $R_1>0$ must hold. Suppose $p(z)$ and $q(z)$ aren't analytic at $z=z_0$, but $(z-z_0)p(z)$ and $(z-z_0)^2 q(z)$ are, then we can find solutions of the form $$ \sum_{n=0}^\infty c_n (z-z_0)^{n+r}, \quad c_n\in\mathbb{C}, $$ where $r$ satisfies $r(r-1)+[zp(z)]_{z=z_0}r+[z^2q(z)]_{z=z_0}=0$. Here $|z-z_0|<R_2$ for some $R_2>0$ must hold. The latter is Frobenius' method.
Now to illustrate my question suppose $z_0=0$ is a singular point of $p(z)$ but not of $zp(z)$ so we would apply Frobenius' method. Then suppose it turns out that $R_2$ is really small. In my understanding we could also have looked for solutions around some analytic point. Suppose $z_0=2$ is analytic, then we would've looked for solutions of the form $\sum_{n=0}^\infty c_n (z-2)^n$ and then maybe the corresponding $R_1$ would have turned out to be more satisfactory.
So if what I said in the above is right, my question is: can we know beforehand if applying Frobenius' method is useful?