In a perfect world, enterprise systems would be designed and implemented around a set of master data with the definitions of the data standardised and carefully controlled.
So says Jane Thomson, executive director at EOH, who defines master data as the data used across organisations in many systems and contexts.
The reality, says Mike Healy, senior consultant, EMEA, at Innovative Systems, is that, in practise, this is rarely the case. Healy says a huge problem exists in organisations with a single customer view; data is often duplicated and not merged, providing an inaccurate view of the information in the database.
Thomson agrees, saying that, in many organisations, there are different definitions of the same customer and this is dependant on what type of system is being written or implemented.
A large part of the problem, according to Healy, is that data quality ownership is not clear within an organisation. Different business departments make different decisions regarding accuracy requirements, he explains, pointing out that operational business units, such as call centres, are normally more conservative about matching duplicate customers than business intelligence (BI) business.
He adds that if a single customer view is implemented, then the operational conservative view of matching duplicates will “trump” the BI view, and this view of the customer will then be fragmented by duplicate customer records - which ultimately leads to poor-quality BI decisions.
ITWeb Business Intelligence Summit and Awards
The ITWeb Business Intelligence Summit and Awards takes place on 28 and 29 February 2012, with a workshop on 1 March. For more information and to reserve your seat, click here.
Inconsistencies are unavoidable in organisations, Thomson says; however, organisations need to look at implementing a master data strategy, which will ensure that there is a designated person in the organisation, who has the power to make decisions and implement them across the organisation. This person would also ensure that each occurrence of data inequality is recorded and cross-mapped, allowing the BI project to cross-reference information, he adds.
Thomson and Healy are speakers at the ITWeb BI Summit, taking place on 28 and 29 February, at The Forum, in Bryanston. The event will host local and international BI industry leaders, with a post-event workshop on 1 March.
At the summit, Thomson will examine why 'bad' data destroys BI projects, while Healy will give an overview of a robust data quality methodology and explore why duplicate customer profiles are still a problem in SA.
For more information about this event, please click here.
Share