I would assess the tool's error handling and logging features as part of it involves having a sampling thing. Depending on the definition of who's doing this, I used to work at a university where we looked at students who registered, and there were people who could register without specifying their gender. Most people know their gender. Some people would not put their birth date, so their date of birth was the default century. In 1989, we had hundreds of 89-year-olds because it was noted as 1900. What we would do is quickly assess that this is not reasonable; in a college setting, how many 89-year-olds are standing in line for freshman orientation? We could produce transformations on the fly. You can look at stuff, but ultimately, you need to go back to the data. There's logic involved; do you have to have a six-digit zip? Do you need the six plus four? It all depends on your criteria. It's doable, but it takes time, and if there's a lot of that, they usually hire people just to handle it. It's a tool that has its limitations; they haven't added wonderful features, and it's a GUI tool based on a desktop that you can run. Their manuals are pretty good, but we're not really using it actively, so I can't tell you everything right now.
