It is like what Ryan said. At some point you have to show that what you are constructing is exotic. This is done, for example, by the signature of a parallelizable manifold bounding your candidate. Such 'coboundaries' don't always exists, however, for homotopy spheres of dimension 4n-1, a large subgroup is of this kind, and these are the easiest spheres to construct. In general, the group of homotopy spheres, $\theta^n$, seats in an exact sequence
$0\to bP_{n+1}\to \theta^n\to \tilde\pi_n$
where $bP_{n+1}$ stands for the subgroup of spheres which bound parallelizeble manifolds and $\tilde\pi_n$ is the $n$-th stable homotopy group of spheres quotiented by the image of the $J$-homomorphism.
Where are you studying these things? Differentiable Manfiolds by Kosinski is a great book. Please, let me know what you are doing, I am always interested in learning and discussing this subject.