What does Bernie Madoff have to do with MDM?

If MDM works on the Madoff Investor List it can work for you!

(1) We mapped the six columns of the registration label to our standard PQI input format.

(2) We ran the jobs in the PQI Package that are executed by the IBM Information Server data integration platform. The processing includes sophisticated standardization, party and relationship discovery. Instances of parties are assigned types including Person and Business. Matching fully leverages the QualityStage® probabilistic matching engine. Instances of person and business at location that describe the same party in the real world are linked together into matched groups and survived. All of the results are written to the PQI Repository.

(3) Then we analyze the baseline metrics that describes the quality and content of one or more party data sets and drill down into the details. In our lab environment the job to process the BMIS investor list completes in 11 minutes.

One important note about developing high quality data integration applications is that it is always an iterative process. That is because every organization's data is different, just as the business rules to process, match and format the data are specific to a particular project implementation. As a pre-packaged application for party data, PQI generates an initial baseline very quickly. From there subject matter experts and technical personnel evaluate the data and confirm requirements and rules with the business. Updates are often made in the source system as well as to the rules for standardization and matching. Then PQI runs again against the input data and updates the repository. Results and deltas are reviewed. The process continues until the output is considered ready for production.

Here are the results from the baseline analysis of the BMIS investor list.

BMIS Investor List Party Quality & Content Metrics
Input rows	13,561
Total party instances	23,076
Total unique parties	13,593
Total unique person parties	7,247
Total unique business parties	6,346
Total discovered parties not primary on any row	5,384
Total number of relationships between parties	11,607
Total person to person relationships	1,963
Total person to business relationships	6,743
Total business to business relationships	2,901

So what does Bernie Madoff have to do with master data management? The registration labels from BMIS serve as an excellent example to connect the dots and explain master data management in concrete terms. Getting master data facts straight is really hard. The problem is pervasive, critical to the business and can get as complicated as a registration label. Solving the problem is a pre-requisite to changing from an account to a true customer orientation. When the process is completed the high quality information necessary to support many critical business decisions is available. The components to implement the party data quality management process and integrate into your environment are here now. It took less than 30 minutes to map the BMIS data to the standard PQI input format and run the PQI jobs. The jobs completed and wrote the results to the PQI Repository in 11 minutes. The logic used and data results generated can be integrated into your environment in a variety of implementation patterns.

If MDM works on the BMIS investor list it can work for you!