Data management: a higher level of thinking

News item -

Tablet and graphs

Banks work with vast volumes of data. It is vital for that data to be uniform and high quality, and usable by the whole organisation. The bank also needs the information to be reliable for it to be used in the mandatory reports that regulators currently require. Webwereld interviewed Frans Gallegos Ruiz and Sandra Wennemers about the importance of speaking the same language.

Over the course of decades, ABN AMRO’s separate departments have developed numerous systems that all use their own data. Then datawarehousing was introduced and the data needed to be compiled, and it became apparent just how much work it is to properly align all this information.

‘Many of the terms used by the various departments and systems aren’t uniform,’ explains data consultant Frans Gallegos Ruiz. ‘Depending on the context, a term might have multiple meanings. At the same time, different terms are used to describe a single concept. Properly mapping out what information the separate processes use requires a great deal consultation with the departments. We analyse the data elements, and can then pinpoint the precise meaning. Next, we draw up a list of clear-cut terms and descriptions.’

Lining up the data

The concept of ‘client’ is a good example of such a concept, data consultant Sandra Wennemers continues. ‘One department might use the Dutch word ‘klant’, while another refers to a ‘client’ or ‘customer’. On the surface everyone seems to be talking about the same thing. If you dig deeper, though, you find the Marketing department also uses it to mean a ‘prospect’: not yet a client, but a potential client. Another department might only use it to refer to existing clients with active accounts.’

So the terms used by the departments contain subtle differences. ‘This isn’t such a bad thing, as long as you’re aware of it,’ Sandra explains. ‘Before you can integrate the data of klanten, clients and customers, you first need to talk to the various departments to establish precisely what data they mean. Otherwise you might find yourself trying to compare apples and oranges.’

‘Besides clarifying the terms and definitions, we also make sure that terms are put in the proper context,’ Frans adds. ‘This makes the precise meaning even clearer and makes associations possible.’ For example a date of birth is associated with an individual, and a planned repayment value belongs to the repayment schedule for a loan.

Crossing departmental lines

Problems rarely arise within the individual departments: everyone there speaks the same language. The challenge comes when people need to communicate across departmental lines, for example with the mandatory reports that regulators require from banks to map out their risks. ‘You need to see the big picture then,’ says Sandra. ‘That’s why we’re compiling a big digital dictionary, to make sure that the whole bank uses the same terminology.’

‘This is making it easier to communicate across departmental lines, because it’s removing much of the static,’ Sandra continues. ‘People understand more easily and more precisely what data is concerned, where the information comes from and how reliable it is. The bank is more responsive and we can answer questions faster. We can manage without the dictionary, but it takes much more time. Data is an asset from which we derive a vast amount of information. A properly organised information setup offers a great deal of added value.’

Centre of Expertise Data Management

This requires the bank to adopt a different way of working, and to achieve this the Centre of Expertise Data Management (CoE DM) has been formed: a team of experienced data specialists who support the organisation, such as Frans and Sandra. ‘We train people, make them aware of the new way of working and the standards that they should use for data management, offer advice and provide resources that they need to complete their processes properly and efficiently,’ Frans explains.

The CoE DM’s data consultants also help with projects where necessary. ‘And if you don’t need our help, we’ll conduct a quality review to make certain that everything is in fact going to plan,’ he continues. ‘The CoE DM uses a book of guidelines, called the DMBOK, or Data Management Body of Knowledge. This provides a framework that describes best practices for every relevant aspect of data management. DMBOK is the industry standard.’

From project relevance to company relevance

The Centre of Expertise is involved in every IT project. Data plays a part in every project, Frans emphasises. ‘We conduct an intake to see where the project interfaces with other areas. What data already exists, what structures can be used to analyse the conditions? Many people are new to the idea that they need to look beyond the project alone. We want them to think at a higher level: we encourage them to see how the data can be used along the entire information chain.’

It is important for people not only to look at the specific conditions for the project on which they are working, but also to wonder whether information can play a role elsewhere in the organisation. ‘This is a shift from project relevance to company relevance. For example, we can explain what data is already managed by other departments, and so eliminate the need for gathering it again.’

Designing a data model

At present the people at the Centre of Expertise Data Management mostly provide support and help with projects. Sandra comments, ‘When a project begins we first take a look at the requirements, which use a particular language. Our first move is to translate the requirements into concepts from the new standard dictionary. This means that everyone is speaking the new universal language as early as possible. Once we’ve translated the requirements, we often also help to design the data model for the database. This way, the bank has more and more of the same data models and our databases are developed according to one and the same standard. A project team can then implement the date model itself. Once everyone is used to this new way of working, project teams will be allowed to design their own data models, and we’ll only conduct the reviews.’

Read more about

Filter on