Garbage In, Garbage Out: Getting Good Data Out of Your BI Systems

Recently, a big topic in BI has been the importance of “data quality” – in essence, cleaning up the data that is presented to the data warehouse so that the errors, inconsistencies, and so on are eliminated as much as possible. What many users do not realize is that this is just a part of a larger problem: How do you get the most out of the business-relevant data floating around, inside and outside the organization? As it happens, I performed a survey and study on the subject last year.

What does this study recommend? The short answer is, you should start thinking of your organization as being in the business of gathering data, turning it into information, and using that information as effectively as possible.  In other words, you think of your organization as trying to get as much high-quality, potentially useful information to your BI solution as possible, and then analyzing that information, followed by using that analysis to make decisions as rapidly as possible. Then you try to develop a set of metrics that will tell you how well you are doing, and what are the weak points in the process.

This set of metrics measure what I call data usefulness. I define data usefulness as the ability to deliver all needed accurate, consistent, and appropriate data to the right user in a timely fashion. My survey in the study convinces me that there are significant and growing problems at every point in the process of converting data into useful information.  Table 1 shows my take on the typical steps in the data-delivery process, the metrics by which the effectiveness of each step should be judged, and the problems that many are seeing today at each step.  The key take-away point is that fixes to one or two steps will not in the long run fix the overall data-usefulness problem. Rather, organizations of all sizes need to take a comprehensive, long-term approach to ensuring data usefulness.

The Data Delivery Cycle

The Data Delivery Cycle also shows that, by users’ own estimation, more than 2/3 of the data that flows into the organization is not used effectively. In fact, if you include the inability to flow new sources of data into BI, more than ¾ of the useful data out there never gets used right.

There is a great deal of more in this study about how to tackle particular parts of the problem, what vendor products and solutions deliver the biggest bang for the buck, and how to monitor the process – material that is far too long to cover here. However, users can take certain steps immediately, without knowing that much about data usefulness. Those steps are:

1. If your company now includes 2 or more (usually acquired) organizations whose operational data is at least partially kept separately, put an EII (Enterprise Information Integration) or “data virtualization” tool in front of your BI, so you can query both sets of data at once. Examples: Composite Software, IBM Information Server.
2. Get a dashboard BI add-on for senior executives, so that you can fine-tune the relevance of the information presented to them, and perhaps be able to display more current information than standard BI reporting tools. Examples: BI solution vendors such as SAP Business Objects and Microstrategy.
3. Make sure you have an Excel option for your BI solution, so useful information can be delivered as spreadsheets as well as hard copy. Examples: Oracle OLAP Option and 1010data.

One caution: today as in the past, the first reaction to problems with data usefulness is to attempt to put everything to do with data in the BI organization, and all data in the BI data store. Experience has shown that this never, or only temporarily, works. There’s just too much data floating around out there, and more new types of data coming all the time. Improving the quality of data coming into the BI solution is one thing; attempting to force everyone to enter all data into the BI solution is quite another. Don’t waste your time and money on the latter.