The counting goes according to the number of independent ``coupling strengths" of the interactions. There is one coupling parameter for each ``simple or abelian" group factor in the full symmetry group representing the interactions. The symmetry group describing the four interactions, with their 4 independent couplings, is (locally) SO(3,1)xSU(3)xSU(2)xU(1), representing gravitation, and the strong and electroweak interactions, respectively. This is the clean way of defining distinct interactions. One meaning of a ``unified theory" is then when some of these symmetry group factors sit inside a larger group, such as the ``grand unified theories" obtained when the three interactions represented by SU(3)xSU(2)xU(1) are a subgroup of a larger group, like SU(5), which has a single coupling parameter associated with it. So the electroweak ``unification" is not a unification in this sense.
So in what sense is it? Prior to the electroweak model, it was not known whether the weak interaction could be described with the same kind of symmetry (gauge) principle as electromagnetism; when it turned out that electromagnetism and the weak interaction were a blend of the two types of interactions collectively called the electroweak interactions, associated with the SU(2)xU(1) symmetry group, people called this a type of unification.
