The more sources you have coming into a stream of customer data, the more likely you are to get some pollution. If everybody just pollutes a little bit, it adds up to a whole lot in the end. Ideally, you want to stop it or catch it at its source.

If you move data quality upstream and embed it in the [business] process, it's much better than trying to catch [flawed data] downstream and then fixing it in all the different applications that used it.