Cooper pairs are pairs of interacting electrons. They do obey bosonic statistics, but they are composite particles extending over many lattice periods, and their internal structure is often important. The same can be said about excitons, phonons, magnons, and other quasiparticle excitations.
Update
Let me outline more precisely in which ways these differences may appear:
- Since cooper pairs extend over many lattice periods, their internal structure is easily probed by external fields, impurities and other available perturbations. This is unlike the case of what is usually meant by bosons, such as laser cooled atoms or elementary particles, where probing the internal structure requires energies of higher orders.
- Another issue is that cooper pairs are actually not bound states of two isolated electrons, but excitations of a many-electron system, like excitons, excitations in Fermi luquid, etc. Thus, they have finite lifetime, and their charge and spin are defined approximately - that is to say that their statistics is close to bosonic, but not quite.
- In presence of many of interacting cooper pairs we cannot really distinguish, which two electrons belong to one pair.
It might be also instructive to compare the BCS wave function with the ground state of the bose-einstein condensate.