Here is an alternative, perhaps more in the spirit of "consider the prime factorization of...".
Let P(n) be the set of (distinct) primes in the factorization of n.
If S is any set of primes {p1, p2, p3, ...}, let F(S) = (1-1/p1) (1-1/p2) (1-1/p3) ...
In this notation, we know that phi(n) = n F(P(n)).
Now it is time for a
Lemma:
If S is a subset of P(n), then n F(S) is an integer.
Proof: each factor in F(S) is of the form
(1 - 1/p) = (p-1)/p
Each distinct prime in S exists (at least once) in the factorization of n, cancelling one denominator in F(S) and leaving
n F(S) = k (p1-1) (p2-1) (p3-1) ..., where S = {p1, p2, p3, ...} and the integer k = n/(p1 p2 p3 ...). The result follows.
Now for the main dish: if m|n, let n = qm. We know that P(n) = P(qm) = P(q) U P(m).
Let D be the set difference P(n) - P(m). We know that D is a subset of P(q).
Therefore, phi(n) = n F(P(n)) = qm F(P(m)) F(D) = phi(m) q F(D).
Since, by the Lemma, q F(D) is an integer, the conclusion follows.