The most frequent data domain addressed in data quality improvement and master data management is parties.
Some of the issues related to parties that keeps on creating difficulties are:
- Party roles
- International diversity
- Real world alignment
Party roles
Party data management is often coined as customer data management or customer data integration (CDI).
Indeed, customers are the lifeblood of any enterprise – also if we refer to those who benefit from our services as citizens, patients, clients or whatever term in use in different industries.
But the full information chain within any organization also includes many other party roles as explained in the post 360° Business Partner View. Some parties are suppliers, channel partners and employees. Some parties play more than one role at the same time.
The classic question “what is a customer?” is of course important to be answered in your master data management and data quality journey. But in my eyes there is lot of things to be solved in party data management that don’t need to wait for the answer to that question which anyway won’t be as simple as cutting the Gordian Knot as said in the post Where is the Business.
International diversity
As discussed in the post The Tower of Babel more and more organizations are met with multi-cultural issues in data quality improvement within party data management.
Whether and when an organization has to deal with international issues is of course dependent on whether and in what degree that organization is domestic or active internationally. Even though in some countries like Switzerland and Belgium having several official languages the multi-cultural topic is mandatory. Typically in large countries companies grows big before looking abroad while in smaller countries, like my home country Denmark, even many fairly small companies must address international issues with data quality.
However, as Karen Lopez recently pondered in the post Data Quality in The Wild, Some Where …, actually everyone, even in the United States, has some international data somewhere looking very strange if not addressed properly.
Real world alignment
I often say that real world alignment, sometimes as opposed to the common definition of data quality as being fit for purpose, is the short cut to getting data quality right related to party master data.
It is however not a straight forward short cut. There are multiple challenges connected with getting your business-to-business (B2B) records aligned with the real world as discussed in the post Single Company View. When it comes to business-to-consumer (B2C) or government-to-citizen (G2C) I think the dear people who sometimes comments on this blog did a fine job on balancing mutating tables and intelligent design in the post Create Table Homo_Sapiens.
Liliendahl on Data Quality 
Descriptions like 360 degree view of the customer, consistent view of the customer and customer data integration reminds me of the search for a singularity. It’s as if we’re looking for data DNA so we can identify each individual by unique characteristics. But even DNA is not unique. If you examine the solutions that provide 360⁰ view, or consistent view you discover there are constraints and limitations. The greater the list of requirements the greater the constraints and limitations. There is a point at which the constraints and limitations reach a point where the solution no longer satisfies all the requirements and so the requirements are compromised.
Terms like 360⁰ view are meaningless. You must define in detail what attributes and characteristics you require for the entity that satisfies the business need. For example do you require eye color, hair color, weight, height etc. In a medical setting these attributes maybe required to provide a 360⁰ view. In the billing department they require the customer’s credit rating and the legal department wants to know if this customer ever pursued a malpractice suit.
Once you define all the attributes required to satisfy all the requirements you face the daunting task of ensuring the quality to meet the diverse needs. Once again the quality dimensions and thresholds vary by context of use. To satisfy all these diverse needs, constraints and limitations of data quality are required.
Once you’ve defined the attributes and data quality requirements with all their incumbent constraints and limitations you may discover you have data unfit for any use. The solution then is to maintain individual data sets in each context of use which is usually the point where we started from.
We see many claims of MDM and CDI success but the details as to the constraints and limitations are rarely published. Of the initial requirements how many were compromised? Of the initial data quality thresholds how many were compromised? A solution may have been deployed but to what degree did it respond to the initial business needs?
I suggest we need ways to measure the effectiveness of MDM, CDI and other data harmonization solutions. We need quantitative data to determine to what extent the solution addressed the initial requirements and expectations. Not statements like “sales went up 20%” since there are many factors that may have contributed to a sales increase. A one-to-one mapping of the initial requirements and expectations including the data required and the quality to the final solution is one way of assessing the effectiveness and success of and MDM project.
Thanks for yet a comprehensive comment here on the blog Richard.
Terms as “360 degrees” and “single version of the truth” are visions we seldom are able reach in reality also because we really can’t define what it exactly is and how it relates to business values.
I agree that there are many factors that may have contributed to a sales increase or other improvements in business related KPI’s and data quality improvement and master data management are only some of the ingredients needed to reach such measurable goals.
Defining, tracking and evaluating data quality KPI’s and examining the impact on business related KPI’s is essential but accurately testing how any one element of our data may affect our business is as fiendishly difficult as accurately testing how any one element of a diet may affect our health, even though we basically know what’s healthy and what isn’t.