Master data management (MDM) can improve the visibility, accuracy and usefulness of critical data, rapidly yielding a return on investment for the organisation. A successful MDM strategy enables the business vision and strategy to become reality.
However, much can go wrong, and when it does it can be a costly exercise for the business, writes Gerhard Botha, principal lead, business intelligence, iSPartners.
Here are the five most common MDM blunders:
No systems integration
The real cost of poor MDM practices tally up when users start analysing the data. When master data from two systems do not share the same names or codes, integration becomes difficult. Disparate sources feed into a data warehouse or reporting platform and numbers become seriously skewed.
For example, a hundred customers from one source plus a hundred from another do not equal two hundred customers, as a customer may reside on both systems and be counted twice. It’s critical to ensure that a master data code like “01” means the same thing across all systems. Failure to do that results in data that can over- or understate the importance of customers.
CRM initiatives will be deemed useless if users cannot populate the CRM system with an accurate golden customer record and don’t have methods in place to ensure data is updated at all touch points.
Often, master data is only made congruent manually – on a spreadsheet. This is the frustration that analysts have when providing data to executives, and it’s why Excel is still the favourite reporting tool in business.
The easiest way to address this problem is the have MDM best practices in place. That will ensure consensus across all systems.
Reporting across systems on master data dimensions can be a nightmare, and the subsequent cost to business intelligence (BI) is high. Again, take the customer as an example: the true value of customers can only be understood if users understand their full basket, as well as all interactions and relationships.
A dimensional business matrix will allow for reporting numbers across sources on the BI platform, but requires effective upstream MDM processes. There are ways of handling master data on the BI platform when poor MDM practices are implemented, but at a cost.
There are three possibilities when loading it into a BI solution if the sources are not conformed:
* Take it or leave it. If the data integrity is dubious then none of the data will be loaded. Thus if any of the source systems’ master data is out of kilter, that data will not be loaded. The problem here is that it may take a while to resolve the master data at source, and users cannot report those numbers in the meantime.
* Master data problems are resolved during the ETL process where exceptions will be managed. Typically, a learning tool can be implemented. As an example, when an exception is seen on the data, such as “zs”, this exception can be converted to “za” by the administrator and will replace all “zses” when they are encountered.
* Simply report anything and everything users get. This will allow data quality issues to be surfaced. However, once users see discrepancies and anomalies in reports, the credibility of the BI solution may be in doubt.
Master data maintenance and lifecycle processes must be replicated in all the systems where the master data resides. This means that if the business process for master data changes in one system, then it must be changed in all systems – that is unlikely to be properly orchestrated.
Difficulty in implementing SOA architecture
Service oriented architecture (SOA) supports business processes by bringing together data from various sources into a single process.
It has long been touted as the solution for retaining competitive edge without the long and expensive process of replacing existing business software, an MDM strategy should support the integration and use of shared data in the organisation, and is one of the key components to any effective SOA.
However, if the organisation finds it difficult to implement SOA, it can become impossible to effectively share services because it is not possible to share the data that is the backbone of the organisation.
Financial auditing of systems is impossible
Data and referential integrity problems prevent organisations from performing financial audits. This is particularly problematic when it comes to legislations like the Consumer Protection Act (CPA).
Organisations will be unable to sell, cross sell and up sell if their systems are not integrated across the business at MDM level, because it is impossible to gain a full view of the customer or of a particular product, which makes reporting impracticable.
These costly problems can be avoided by implementing MDM best practices, which are becoming increasingly important as businesses realise the link between consistent data and business performance. That requires a holistic approach including people, processes, technology and data.