James Taylor's Decision Management

How do you tell if your organization is ready for decision management? If not, what can you do to start laying the groundwork?

Second, Data readiness

In order to derive insight from data you must first have data available. The accuracy, completeness and integration of the data you have will determine the quality of insight that can be derived. Of these by far the most important is the accuracy of data. Building predictive analytic models from inaccurate data or mining inaccurate data for business rules will be fatal - the results will build on the inaccuracies of the data and product inaccurate predictions.

Accurate data, even if limited in scope or completeness, can be successfully used to deliver analytic insight. Clearly, however, the more complete the data the more robust the analytic models are likely to be. Similarly, customer-centric data is going to provide more opportunities for more interesting analytics than account-level data. Thus projects to improve master data management and customer data integration will materially improve the quality of models being developed as well as offering new opportunities for models. Organizations should consider these kinds of projects in parallel with the development of EDM.

Projects to improve data quality and completeness are synergistic with EDM projects. By focusing data integration and quality effort on projects where analytics are being developed it is possible to apply resources more effectively. Knowing what kind of insight is being sought can help prevent wasted effort by focusing on those areas where more or better data will drive better business results.

Constantly evaluating the data that might help make a decision more effectively should become part of the company culture. This might result in projects to enhance data capture, to capture new information about customers or products say, or attempts to find data outside the organization. Some organizations are sophisticated consumers of external data, such as geographic data, consumer data or demographic/census data. Others are not. This kind of external data, if it can be integrated with internal records and applied to specific problems, can add a lot of value to analytic models being developed. As EDM projects progress teams should constantly ask if there is external data that can be applied to improve the quality of insight or to derive new insight. While privacy issues are real and must be properly considered, external data can enrich analytic models very significantly.

Technorati Tags: analytics, data mining, data warehouse, decision automation, innovation, readiness, data