Let $I=(a,b)$ an open bounded interval. It is well known that $W^{1,p}(I)\subset C(I)$. It easy to see that there are $f\in C(I)$ such that $f\notin W^{1,p}(I)$ It is enough to take $I=(0,1)$ and $f(x)=x^{-\alpha}$ for certain $\alpha>0$ depending on $p$.
We are trying to find $f\in C(I)$ bounded such that $f\notin W^{1,p}(I)$. If $f$ has a countable many (even as many as a null set (?)) peaks, surely $f\in W^{1,p}(I)$.
Our first guess is that the Weierstrass function (continuous but non-differentiable everywhere) could work, but we want to know if there is some standard (easier) (counter)-example.
Since any function in $W^{1,p}(I)$ equals the integral of its weak derivative, which is in $L^p(I)$ and hence $L^1(I)$, it is absolutely continuous. Hence any function which is bounded and continuous but not absolutely continuous will be a counterexample. For instance, the Cantor staircase function.