Use that $\int_1^{\infty}\frac{1}{x^p}\,dx$ converges if and only if $p>1$ and diverges for $p\leq 1$ (this is easy to prove in case you're not familar).
When $k\leq 1$
$$\frac{\log(1+x)}{x^k}\geq \frac{\log(2)}{x^k}\;\text{for}\; x\geq 1 $$
and the divergence of $\log(2)\int_1^{\infty}\frac{1}{x^k}\,dx$ implies the divergence of the given integral.
When $k>1$, there exists an $A>1$ such that $\log(1+x)<x^{(k-1)/2}$ for $x>A$. Hence,
$$0<\frac{\log(1+x)}{x^k}<\frac{x^{(k-1)/2}}{x^k}=\frac{1}{x^{(k+1)/2}}\;\text{for}\;x>A $$
and the convergence of $\int_A^{\infty}\dfrac{1}{x^{(k+1)/2}}\,dx$ implies the convergence of the given integral (note $(k+1)/2>1$).
What you've done with integration by parts is also a valid way to prove convergence in the case $k>1$. You have to show that both the limit and the second integral is finite (which basically reduces to the fact I cited above).