We make substitutions in order to make integrations simpler. It is therefore useful to make substitutions that bring together terms inside functions. For example: when integrating f(x)= cos(3x -1) it would be useful to make the substitution u = 3x-1 making the integration cos(u).Another thing to look for is whether a substitution can be differentiated to be equal to a term in the integrand, as when you change the integration measure it will eliminate that term. For example in the function f(x)= cos(x)sin2(x) it would be useful to make the substitution u = sin(x). When this is differentiated to find du it gives du = cos(x)dx. This eliminates the cosine term leaving an integral over u2.