I was having interesting coversation about MDM with one of my customer last week. It's mid sized bank and they are in process of evaluating MDM. When i asked him " How confident are you about quality of your data?" He said "Honestly, I do not know". Then i told him that for MDM one of prerequisite is to have good quality of data. If quality of your data is not good then MDM solution implementation is bound to fail.
I have seen quite a few organizations who would like to embark on MDM initiative without having good data quality system in place. Thanks to huge amount of marketing money spent by some of large IT product vendors. Quite often organizations fall in this trap and end up investing hugh amount of money and efforts.
There are two types of MDM solutions in market. Operational MDM and Analytical MDM.
Operational MDM is used to collect customer information at front desk. This solution is used to standardize the mechanism to capture customer information at various customer touch points in organization. Typically, organizations have 5-20 customer touch points. This solution provides customer information to various operational systems in organization.It ensures that any changes made in customer information at any of customer touch point are transferred to all operational systems. It will work in organizations which are still in process of implementing operational systems and have very few customer touch points. This approach requires discipline, and huge amount of training efforts. Currently most of banks in India have operational systems in place. These operational systems are built using old technology and captures customer information specific to their application.Enhancements to these oprational systems are very time consuming and lead to performance issues. Hence this solution will not be suitable for most of the large and mid sized organizations. It also requires huge amount of training efforts to train front desk staff on this solution.
Analytical MDM is used for historical and predictive analysis. This solution sources the data from transactional systems such as CRM, ERP, CBS, LOS etc...Analytical CRM can be updated once in a day or multiple times in a day. I have seen banks updating it once a day which is more than sufficient to cater to their current business requirements. This solution doesn't require retraining of front desk staff. However, it requires tight integration with transactional systems. Analytical MDM should be SOA enabled. This will enable source system to call web service and check whether new customer is already customer of bank or not. Analytical MDM will also provide information related to class of customer(Preffered, Gold etc) and behaviour based on past transactions. This will help to take decision about loan approval or issuing credit card or giving prefferential service to your customer.
The way Analytical MDM & operational MDM store the data is also different. Analytical MDM stores data in denormalized format so that it can be retrieved easily for analysis whereas Operational MDM stores the data in normalized format so that it can be updated quickly.
Operational MDM stores demograhic details such as age, birthdate, name, address whereas Analytical MDM stores information related to profibility, behaviour score, credit score and propensity to buy product apart from demographic details of a customer.
Both types of MDM solutions require strong data quality engine in backend. This data quality engine should be capable enough to address peculiarities in Indian addresses and names.