For decades, we have relied upon data integration approach to setting up master data management systems. While there is nothing unruly about the approach itself, the perception it creates on the implementation team has been troubling to say the least. MDM Implementation

I wrote an article here about approaching MDM in a holistic and agile fashion. I debated here about taking a business driven approach to be successful with MDM. The focus ought to be on identifying the data issues and aligning MDM with data governance to bring master data to a common, agreeable, enterprise standard structure. Data integration then just becomes a way to move the data, clean it and maintain it in a place which will act as single version of truth.

Given that quality data creation is your primary goal, following 3 aspects become the key drivers of an MDM implementation. Unfortunately, these are also the 3 most under looked features as I have experienced over the years.

Data Discovery and Profiling

What happens when you (Master Data) visit a doctor (MDM System)? The first thing your doctor wants to know is to see what your health history and symptoms (Data Quality) looks like. This is so he can find the perfect treatment (Transformation) for you and improve the ‘quality’ of your health.

How do you feel if your doctor never checks your symptoms before giving you a medicine? Data discovery is very important for MDM

Same analogy applies to MDM. Master Data suffer from quality issues such as non-standard representation, in-consistencies, missing and defaulted values. As you bring this data into MDM system, you have to run it across a systematic discovery process where anomalies are identified and documented. A profiling step ensures data is tested against rules and fixed.

I wrote about this step in my earlier blog on five key factors in architecting a master data solution. Missing this step is sure sign of failure and will lead to data quality issues getting replicated to MDM system. How do you feel if your doctor never checks your symptoms before giving you a medicine?

Enriching MDM with External Data Services

In the past, many systems and business processes have relied on external data sources while making their key decisions. Sometimes adding additional value and many times part of a compliance procedure, these external sources have proved to be of significant value. They have helped in making marketing and sales team take proper action based on intelligence available outside organization walls.

Social and other big data sources can enrich your master data and help make it more relevant and current.

Same principle applies to MDM system. In fact, MDM allows a well-coordinated re-use of external data. This helps in building complete profile of your master data records in a central place, so all your applications can leverage this intelligence.

Added to this is the availability of social and other big data sources that can further make your master data more relevant and current. A well-planned enrichment process from commercially available data providers can help you make easier segmentation and prospecting of your master data. This should be a key aspect of your MDM journey as it helps in areas such as – address standardization and certification, social data enrichment, data quality improvement and real world alignment.

Reference Data Management

One of the most painful parts of the MDM implementation is when customers fail to distinguish reference data from master data. There are few reasons why this is a significant issue. One, reference data requires a focused and dedicated effort in itself to manage it well. Two, reference data generally gains more value when it is widely re-used and referenced. Three, failure to handle this data and standardizing it will cause complexities to MDM implementation itself.

Reference data requires a focused and dedicated effort in itself to manage it efficiently

I have seen customers building temporary staging areas which are a huge pain to maintain for the implementation team. They also cause taxing issues in terms of accountability. Result of all of this is that you will have – dropped records, duplicated values and overhead of manually maintaining mappings for individual sources. While on this topic, do visit my earlier blogs to read about reference data management.

MDM implementations continue to be characterized by lengthy timelines, effort and obscurities. Knowing the best practices and learning from experience is one of the best ways to ensure you make your project fail proof. I hope the three aspects I discussed here will help you plan your MDM journey well.

3 cheers to your success!! Please provide your feedback via comments. I love to hear from you.

Image courtesy of John Kasawa/sippakorn/