Not being able to find the right inventory or products, because of wrongfully assigned categories, missing descriptions and details, can cost businesses a lot of money. If a customer is unable to browse laptops, because they are filed under notebooks, will make you loose the business. If employees cannot find spare parts for the same reason and reorder, you'll find yourself with inventory that is there, but nobody knows about it. You basically put your money on the shelve.
Master data is in a constant state of deterioration. New data is entered, data is imported or merged with other datasets, new details for products are emerging etc. While the importance of cleaning master data is undisputed, it is also a pretty tedious and lonely task for whoever poor folk who has the task to do it. Hundreds of thousands or records require full concentration, quite some knowledge about the what the data means, and in the end, nobody will notice good master data. It's like with manners: nobody notices them, until they are bad.