Yesterday I went to the MFA in Boston, and found my way around the museum with the map. The architecture is a little bit complicated: for example, because of some atriums (atria?) that span several levels, sometimes the best way to go from one room to another room on the same floor is to go down several levels, across the atrium, and back up several levels. That makes the map design challenging. (In particular is it not clear when what appears to be a "room" on the map actually corresponds to some empty air floating above a floor at a lower level.)
As discussed in a previous thread, to help the visitors orient themselves quickly, they partition each floor into sections (clusters of rooms) that have different colors, depending on the theme of the art displayed. But to my dismay, the correspondence is not a bijection: although the colors correspond exactly to types of art, art of the same type may be displayed on several floors.
As a result, before I started my visit, I had to spend a good while pouring over the description of the rooms on one page, and the map on another page. Very inefficient! And it is not until after I had left that I noticed that they had highlighted a few particular pieces of art as particularly noteworthy: I was so busy trying to understand their visualization system that I did not even notice those marks.
All in all, a less than optimal example of data visualization.