All of us depend on data created elsewhere to do our work. In the face of errors, most people’s natural reaction is to correct such errors in the data they need — after all, when you’re dealing with a mountain of day-in, day-out demands, that seems the fastest, most efficient way to complete the task at hand. The problem is that finding and fixing flawed data soon becomes a permanent fixture. Writ large, it is expensive and time-consuming. Worst of all, it doesn’t work well: Too many errors leak through, rearing their ugly heads later on and leading to larger mistakes, bad decisions, and angry customers.