For these kinds of exam problems, I recommend to use your intuition. Once you understood the difference between the two formulas, you should be able to express that difference in terms of a transition system.
If you don't have that intuition, one way to apprach a difference is to consider the recursion of the temporal logic formulas. In the mentioned case, we get
AF(p∨q) = (p∨q) ∨ AXAF(p∨q) = p ∨ q ∨ AXAF(p∨q)
(AF p)∨(AF q) = (p ∨ AXAF p)∨(q ∨ AXAF q) = p ∨ q ∨ (AXAF p) ∨ (AXAF q)
Hence, both formulas are satisfied if p ∨ q holds, so we have to make both false for a counterexample. Then, you can satisfy the first formula if some successor states satisfy p, but not q, and the remaining successor states satisfy q but not p. Thus, you need an initial state where p and q are false and the two successor states with self-loops.
The above idea with the recursion usually works fine.