Better management of the stable non-transactional data that organisations store and manage can be good for business, cutting unnecessary costs, improving efficiency, and enhancing decision-making – and not just in obvious areas such as customers, suppliers and products. The finance function is also important, as FSN writer Lesley Meall discovers when she looks at ‘master data management’.
Let’s start by getting personal. I moved into my current home seven years ago, since when, the gas provider has been sending all fuel bills addressed to me. But a couple of years back, when I took out a boiler maintenance contract with the same provider, it started sending the related correspondence to the previous householder. I have pointed out the discrepancy, twice, but to no avail. As a contributor to a well-known finance magazine I am on the mailing list (in various guises), not once, not twice, but three times. I have pointed out this duplication, but to no avail.
I could go on; I expect you could go on yourself, actually. Who doesn’t have their own similar tale of woe? Even the smallest organisation now holds a sizeable and growing amount of ‘stable’ non-transactional ‘master data’ on people (such as customers, employees, and vendors), places (such as branches, cost centres, delivery addresses, stores, and revenue centres), things (such as products, parts, and assets), and concepts (such as contracts, profit centres, and the fiscal calendar), and often within a number of different databases. So managing all of this can be complex and challenging, and not managing it can be costly and counterproductive.
But if you do want to address any of the very many associated data management issues, where do you begin? It’s probably a good idea to start by clarifying what computing professionals are on about, if and when, they refer to master data management (MDM). Those who already know will probably want to find something better to do with their time, while the rest of us try not to be over-awed by the all encompassing nature of MDM, as it comprises the very many processes and tools that can be used to collect, aggregate, match, consolidate, and quality-assure ‘master data’ to ensure that it’s consistent.
As Google can very ably demonstrate, MDM is a vast and complex area. But at its simplest, it’s about trying to deliver to users of data, that Holy Grail of Holy Grails: a single version of the truth. But if we put this phrase aside (because some people really don’t like it), MDM is about trying to ensure that an organisation doesn’t use multiple, and potentially inconsistent versions of the same supposedly ‘stable’ data in different parts of it’s operations, or even hold duplicates of almost identical or even incorrect data in one or more databases – which can result in the sort of pointless confusion (and irritation) and wasted resources I mentioned above.
But the all-encompassing nature of MDM means that in practice, it means different things to different people. Vast, wide-scale projects tend to be the preserve of very large organisations (on which more, later), where MDM may mean using data governance mechanisms (such as workflows) to enforce policies that can improve data quality and consistency, and reduce data complexity across multiple systems, operational units or enterprises. At the other end of the spectrum, it could mean capturing, verifying or batch-cleansing addresses, or enhancing them with data such as telephone numbers or lifestyle, to support marketing initiatives.
The need to get this type of data correct links organisations of all shapes and sizes, as does the potentially high cost of getting it wrong. According to the UK carrier Royal Mail, a company that posts 200,000 items only needs an error rate of five per cent to waste over £100,000 a year – and that relates just to the direct costs involved in the mailing and its content. When PwC did a global survey of data management, it found that 75 per cent of Australian, American and English companies experienced significant problems as a result of defective data, with a whopping 30 per cent failing to bill or collect receivables as a result.
As there are lots of types of systems designed to address these (and other MDM) issues, the route you take to improving your master data will depend on numerous factors, ranging from the information systems you have in place, through how much you can afford to invest, to what you are trying to achieve – and the unique problems associated with individual scenarios. At Graphic Data UK, for example, the focus is on capturing and validating source data quickly and accurately, but this isn’t easy, as it provides this as an outsourced service, and processes 2.5m credit card applications each year from Austria, France, Germany, Spain, the Netherlands and the UK.
MDM & customers
So the business uses a range of data management tools. One of these is Matchcode International, from Capscan: a global address management solution that allows the 30 employees at Graphic Data UK to validate, correct, standardise, and format address data for all of these regions (and it can do so for over 240 territories worldwide). ‘It has minimised the number of keystrokes the data entry team needs to make to identify a full, valid and accurate address,’ says Andy Boniface, group project manager at Graphic Data UK, and you might be surprised by how much. ‘Instead of around 50 key strokes, they now need to input just seven or eight,’ he reports.
The system also works in conjunction with another data verification system, which addresses international bank details such as sort codes and account numbers, so that Graphic Data UK can automate as much as possible of the validation process, and ensure its accuracy, prior to submitting it to the credit card company’s database, preventing a number of potentially costly problems. ‘Correct addresses are essential for the prevention of fraud such as identity theft,’ he says, and to ensure the correct credit scores are applied, and as he observes: ‘Incorrect data is incredibly costly to resolve.’
At Berlingske Media, one of Denmark’s largest media groups, a wide-scale MDM implementation is under way, using a complete MDM solution, from Talend. The media group produces content for online, web-TV, radio, mobile phones, and print. It’s the parent company of 11 national and regional daily newspapers and more than 50 local newspapers, and its printing works service the group and outside clients, which adds up to more than two and a half million customers, so the MDM project is complex. ‘We expect the MDM to efficiently integrate all of our data, whilst helping us aggregate and organise our customer data,’ says Torben Lundberg, the CIO.
But he is under no illusions about the scale of the undertaking. ‘We have set ourselves very ambitious goals withour customer centric programme,’says Lundberg. Talend MDM will be used to manage the entire process from profiling and quality to integration, modelling, mastering and workflow. ‘It’s essential that it makes us able to share customer data across our entire business, secure data quality, and cross analyse and segment it easily and accurately,’ he says. ‘It’s essential that we have the flexibility to grow the initiative from a small scale before involving all of our data, ensuring that our master hub is completely accurate.’
MDM & finance
Although customers are the element that most people think of in the context of MDM, projects involving suppliers and products can also deliver significant benefits, and for some organisations – particularly those with international operations – financial master data also looms large. ‘At Intel, we have been building our MDM systems for the past seven years,’ says Vanitha Srinivasan, an enterprise architect with the technology giant, and Intel has built a system that is a mixture of off-the-shelf MDM products and a home-grown MDM system, based on workflow.
Finance was one of the first areas to get a central master data solution, and this has been in place since the 1990’s, but it has needed to evolve, to reflect changes such as the growth of the business, the evolution of finance systems, the development of financial reporting standards, and the transition from national Generally Accepted Accounting Principles (GAAP) towards International Financial Reporting Standards (IFRS). ‘Managing finance master data poses distinct challenges,’ says Srinivasan, as accounting controls, regulations and reporting standards change over time, and this has to be reflected in processes and applications.
‘At Intel, finance master data management has a two-fold scope, encompassing organisational master data and finance master data,’ she explains. The former refers to profit centres, and cost centres, and the latter to different types of currency, general ledger accounts and the fiscal calendar. But whilst a certain amount of finance MDM (FMDM) can be automated and enforced with the workflow that Intel has built its home-grown MDM system around, the governance of key finance master data, policies and definitions, has to be carefully overseen by a central finance data maintenance group.
‘Governance is a major factor in maintaining data quality,’ she says, and the group is responsible for a range of activities including the configuration and validation of bank master data and currency codes, syncronisation of business and operations hierarchies, and audits on data owners (to identify who uses master finance data and who authorised access to it). ‘This centralised group gives us control of data, but it’s not ideal,’ she admits, because of the fragmentation caused by multiple business teams being involved in the approval of maintanance decisions – and the current FMDM system can handle only one chart of accounts.
‘The finance business unit is currently evaluating the implications of moving from US GAAP to IFRS,’ says Srinivasan, and the architecture for managing finance master data has to be particularly flexible, in order to accommodate changing standards. It remains to be seen if Intel will create this flexibility by adding multiple chart of accounts to its existing application, or opting for one of the specialist products on the market. ‘Due dilgence is essential to selecting the correct MDM product,’ she says, but in terms of managing master data, ‘governance is the key to continued success.’
