MDM the foundation for a complex system architecture

Ville1_web The challenge for many companies is a complex system architecture, where systems communicate poorly with each other and the quality of information varies. This is where Master Data Management (M|DM) can be introduced to the core of a company’s data management processes to provide essential stability as well as other surprising benefits. Ville Koivumäki, development manager at Lassila & Tikanoja shares his views of the importance and implementation of MDM.

Koivumäki leads the BI and data management architect team, and has had a central role in taking MDM to the core of IT-development at L&T. The company operates in four countries and employs 8000 people. L&T has grown significantly through acquisitions. Different IT cultures and system languages have been incorporated into the company over the years.

”The importance of master data is usually realised through reporting,” says Koivumäki, speaking with experience of three separate ERP projects. “If you enter poor-quality data in the beginning, you cannot expect diamonds at the end. Therefore, it is crucial to pay attention to the information source, and to create clear and controllable models for entering master data. This is why MDM has become a key area under my responsibility.

Information management – the foundation for new business

MDM is particularly useful when the company has several ERP systems and it is difficult to obtain visibility of customer data. One of the early stages of an MDM project is to define what information is crucial for business operations and who, for example, the customer actually is. During the project, the number of terms is typically narrowed down to a few dozen core concepts. This process involves long conversations during, and even after the project. In the world of digitalisation proper MDM generates the capability to create new businesses and e-services. Koivumäki credits the management of L&T for their experience and support, which he says have been highly important.

L&T had no previous experience of master data projects and even the discussion around the concept of MDM was new, to some extent. Furthermore, doubts were circulating due to the awareness of failed projects in other companies. Lassila & Tikanoja started to seek a new, alternative model for their MDM solution.

Impact on information architecture

Information architecture defines the needs for MDM, but the issue of best practices should also not be ignored.

In addition to defining the core concepts, it must be fully understood where data is used and how it is delivered to the user. Some data is integrated and managed jointly, but some will have a local presence.

In their vision Lassila & Tikanoja handles only their own data; information which they have the knowledge and ability to process. Data which they cannot affect, such as a customer’s postal information, is purchased from external sources. In the future they wish to rely even more on external data, made automatically available. Even today, when the Finnish Postal Service updates their address register, 35 000 people in 15 000 companies in Finland update the modified data to their own registers. This makes no sense.

In discussions about data automation, it is not just external providers that are relevant, but also the development of connection between customers and suppliers. Additionally, there is the question of internal automation, which reduces work and improves quality.

Impact on application architecture

L&T has over 160 business applications, including Excel software. The number of most relevant business applications is around 30-40. With such fragmented architecture, MDM becomes critical.

One traditional example of implementation is the opening process of a customer to an ERP system. Most systems require a separate client opening process, despite this not being in accordance with normal business operations. This was something the L&T MDM pilot project totally turned on its head! The flow of a typical customer call was analysed to find out in which order information is acquired. It turned out that only the last questions concerned the name and address. Now MDM is integrated in the process it gives L&T the ability to distribute information to each system. The work is now business and process driven, not application driven.

Even a complicated application framework can be taken over with uniform master data and effective integration. A flexible application architecture is the biggest added value of the MDM-model according to Koivumäki. New applications can be easily removed and added when master data locates in the MDM and not within separate applications. The cost and time of modifications declined dramatically.

However, MDM doesn’t always have to be the only source of master data. If information is naturally generated somewhere else (e.g. in an HR system, which is the master of employee information) the solution must support this. This way the customer can decide which data is managed centrally and which from separate systems. The hybrid model is excellent, as well as the fact that there can be different governance models for different concepts running simultaneously. Different governance models can even be applied within concepts at the field level.

”The world is changing and systems evolve, so it makes no sense to set the system framework in stone. With the help of MDM it is much easier to deploy new systems. When the system is ready, qualified and uniform master data can be downloaded from MDM on the very first day of use. Then, everything is ready for transactions.”

Impact on integration architecture

The interpretation between the systems used in the integration architecture can also be built into the MDM system. This means that MDM is able to recognize the different data formats that each system requires for data distribution. Integration tools are just “dummy” data transferring tools, which do not alter the format of data between the systems.

”From the information management point of view there is a big difference whether interpretations take place between each system or centrally in the MDM system. We at L&T found it more rational to have data centralized in one location, where it is easier to control.”

Small but productive steps

”It is reasonable to start with a small and concrete step. The approach is much lighter than an MDM project which would encompass the whole IT-architecture and cost millions, and in the worst case it could fail after years of effort.”

The first master data projects were introduced to the management as rationalisation projects that would achieve measurable financial benefits. The improved data quality is a bonus. It is a common mistake to try to justify the project to management using soft, non-measurable values. The right way to do it was to find a business case that translates directly into financial value.

The long term objective of the MDM-based development work is agility. Instead of a massive project taking many years, systems are seized one at a time. The system architecture is constantly evolving and master data must be able to connect to new systems at any time. Systems should need little or no modifications and should be harnessed as such to produce better information and to operate better together.