The specific examples of fraud Jürgen discusses is the misdeclaration of values or origin for textiles, which is an evasion of ad valorem duties. A scatterplot of commodities versus time helps visually cluster cases that meet certain criteria that imply fraudulent activity. Interestingly, the clusters will often - and suddenly - alternate between neighboring countries, depending on which is known to be enforcing customs regulations at the time. In his office, Jürgen proudly displays a wall-sized plotter printout of the scatterplot across all nations.
My favorite part of the presentation came when Jürgen described a request he heard from Ellie Fields, our Director of Marketing: could you show how this was visualized before Tableau? Jürgen smiled as he showed a blank PowerPoint slide, and remarked "I didn't even try - it would be ridiculous!"