Interesting feature by Julia Turner at Slate on why hand-drawn maps are sometimes more useful than those generated by Google or Mapquest.
In archaeology, I find hand-drawn sketch maps extremely useful. A sketch like the one below, by my former colleague Lisa Kraus, provides more information about a site than you would get from even a very detailed topographic survey. A trained eye can pick out things about features that are barely present in the physical data; I have often been surprised to find that a feature that looks perfectly round to my eye shows up in a topographic map as a couple of lumps in a vague trapezoid. An experienced archaeologist can spot patterns within the noise, seeing rectangular foundations within piles of rubble, or patterns of ditches and banks so subtle they would only show up on a topographic map with 1-centimeter topo lines.
On the other hand there are things for which sketch maps are not accurate, like gauging the distances and alignments of widely separated features. For that sort of thing, an accurate survey as essential. So ideally you would have both an overall site plan based on a real survey or an aerial photograph and a sketch showing the interpretations of the archaeologist.