I love wine tastings. I listen to an expert tell me all about a certain vintage’s bouquet, balance, smoothness and finish. Then I taste it and I either like it or I don’t; though I do know that fine wine usually doesn’t come in a box; even if I like it.
Like wine, data has specific criteria by which to judge its quality. Generally speaking, those criteria include accuracy, completeness, consistency, uniqueness and timeliness. Too often, though, folks are willing to hear all about what makes up high quality data, but then opine that their data is just fine for their use even though it is really like wine in a box. They may like it, but it really isn’t good enough.
But honestly, who really cares? Well in the case of data quality, the stakes can be so enormous that you should. Let’s say your company has been a long time supplier to Time Warner. The company’s name is entered into different databases as Time Warner, Time-Warner, Time-Warner Inc., AOL/Time Warner and TWX (the Time Warner stock symbol). Generating a graph showing your total business with Time Warner could be problematic.
Some may see that example as pretty small potatoes and, under certain circumstances, not very hard to fix. The most stunning example of the potential impact of poor data quality is the 2000
The 2000 presidential election seems like an extreme example but it really isn’t. Anybody following the crisis in
So does poor data quality actually matter? Well the history of the world–and the future of your company– could turn on it.