Thursday, April 3, 2008

Data Warehousing: Siloed Data--Part II

This continues our definition of a data warehouse and our discussion of why you need one.

The second characteristic of your operational systems is that the data is typically "siloed." This is lingo meant to illustrate that the data in each system is independent of other systems. For example, the data in your sales system might be totally separate from the data in your financial system. But in real life, nothing could be further from the truth. This data is inextricably interdependent and you may have an analyst whose full time job is pulling revenue and cost data to analyze your most profitable market segments.

The problem with siloed data, aside from the fact that it often needs to be queried from multiple systems and "massaged" to produce meaningful analysis, is that often the same information is manually entered in multiple locations, inviting inconsistencies and inaccuracies. But, even more importantly, the data isn't always defined the same way in different systems, which leads to discrepancies in reporting. Sometimes you may not even be aware these differences exist without intensive investigation, but often you will be met with situations (as an example) where you feel you met an important goal, but your boss thinks you fell short. See "Not My Incentive!" for further instructions.

(to be continued)

No comments: