Blog: Jill Dyche

« Having The Right Gear | Main | Zen at Work »

Dim Sum and Then Some

In which Jill goes in for Europe and comes out with a steaming bowlful of Asia. And some kick-ass shu mai.

So I’m on my way to speak at a conference in Asia and I have a 3-hour layover in Taipei. I make my way to the Dynasty Lounge to plug in, turn on, and scarf down the customary shortbread biscuit. I’ve found that shortbread biscuits are a staple in business class lounges worldwide, so it would be wrong not to have one. Or two.

When I enter the lounge, my expectations are exceeded: the lounge has a Dim Sum bar. Spread out are all sorts of Chinese delicacies from little steamed dumplings to plump meat pies to piping hot noodle soup with meatballs. It’s a veritable cavalcade of Asian cuisine! I grab a pair of chopsticks and a spoon, ladle some soup into a bowl, load on some chili sauce, and dig in.

This is how companies feel right after they finally get to buy a data quality tool. They are delighted with the number of options they have, but at the same time conflicted. After all, consider the choices! Where to begin?

Many of our clients begin the automation of data cleansing with a subset of customer data. This is largely because customer data has the most management support, and receives the lion’s share of executive attention. If cleaning up customer data means generating more accurate predictive models, which in turn drive higher marketing hit rates, most managers will happily jump on the bandwagon.

Other companies focus more on business needs. A client of ours that recently re-engineered its supply chain started its data cleansing with product item data, planning a horizontal expansion to suppliers, then eventually to customers. It’s a requirements-driven approach, and there’s nothing wrong with that.

The point is, there’s no one right answer, but the old adage of “start small, think big” is an apt one here. Viewing a new data quality tool as a black box that will process all the data on the data warehouse like some sort of meat grinder—in goes the data steak, out comes the spicy, Chinese meatballs delicately flavored with cumin and chili, oops, sorry—can do more harm than good.

A deliberate approach to prioritizing data cleansing projects will serve you well. Understand which data subject areas or subsets will yield the highest improvements or drive the best business decisions, then chunk up your data cleansing accordingly. There’s plenty of data to go around, and you can always go back for seconds!

Technorati tags: data quality, data cleansing, data governance, data warehouse, data management

Share: