From Marian Petre's recent paper "Mental imagery and software visualization in high-performance software development teams
...even in debugging and comprehension, the experts relied more on their own systematic practices than on visualizations---and their use of available visualizations related to how directly the visualizations supported their practices. Tools which simply re-presented available information, which failed to provide forms of abstraction or selection, or which embodied assumptions at odds with the experts' practices were discarded.
Take-up was extremely low in the context of design and generation. The exception was for general tools such as MATLAB, which allowed software developers to realize their own visualizations---in effect for visualization-builders rather than visualizations per se...(Please note that 'Not invented here' was never offered as a reason not to use a tool.)