Digital Transformation in chemical research and development

Kundenprofil:

Chemical Industry
Approx. 3 billion € revenue

Together with our customer, we developed a comprehensive concept in a period of 3 months to implement and establish smart data-driven solutions in the area of research & development. During the needs analysis we conducted, it became clear that extensive organizational requirements must first be created in order to be able to address data science use cases efficiently and sustainably in the future. Together with the customer, we therefore developed a concept for the future IT landscape of research & development, which should make identified potentials usable with little effort (‘low hanging fruits’) and at the same time act as an enabler for future data science implementation.

SITUATION

In view of a very agile and highly competitive market, the rapid and successful development of new products for our customer is essential for commercial success. However, it is also a common characteristic of the research and development business area to generate a high level of costs without any guarantee that it leads to development of new, groundbreaking products that can be sold on a market. 

In order to make R&D more efficient and reliable, our client considered the extent to which they could use Data Science for this purpose. Based on our expertise, we were asked to evaluate already identified use cases as well as to work out further promising use cases together with the customer. The initial plan consisted of sketching a first prototype for a use case and providing it with a possible development plan. 

ANALYSIS & CHALLENGE

To better understand the needs of the users, we first conducted a series of on-site interviews with the various stakeholders. The goal was to identify challenges and unused potential as well as to take a closer look at the IT system landscape. The insights and processes gained from this were then collaboratively deepened in several half-day workshops with the respective stakeholders and recorded on a digital whiteboard before we condensed and processed the results and coordinated and discussed them with the various IT departments. 

It became clear early on that Data Science alone would not be the solution for the future success of our client’s research & development. A number of organizational, process-related and technical hurdles were identified that had to be overcome before we could recommend a Data Science implementation without reservation. For example, the customer was confronted with a very heterogeneous research IT landscape, which had developed largely independently of that of business IT due to the special requirements of research & development. 

In short, a multitude of local, non-synchronized data sources, different workflows, a wide range of applications in use, and different chemical data models with varying granularity made it clear that this would not be the best prerequisite for Data Science. 

We therefore recommended to the customer that, as a first step, the basic technical and organizational prerequisites for a Data Science Platform should be created before Data Science itself should be driven forward more intensively. Our recommendation met with a positive response from the colleagues concerned, so that the focus was now no longer on the use cases themselves, but on the concept and a roadmap for the future of research IT.

IMPLEMENTATION

Based on these findings, we worked with the customer to develop a fine-grained concept for transforming the existing IT landscape so that data science use cases can be implemented efficiently, cost-effectively and sustainably in the future. At the same time, the concept also included solutions for the problems and concerns of the employees, so that they can, for example, access the right data, at the right time, in the right place. This ensures that no more time is wasted trying to locate required reports in one of the many data sinks. 

A central component of the concept developed was the establishment of a single point of truth in the form of a research-owned data warehouse (DWH) with connections to all important data sources and applications. Subsequently, a data lake was to be set up parallel to the data warehouse, which would allow the collection of semi-structured and unstructured data. In addition, it should be possible for researchers, developers, analysts and data scientists to flexibly tap into both the DWH and the data lake via an interface. 

Apart from the technical aspect, we considered it enormously important to also prepare the research and development organization for the upcoming transformation – after all, technology is only beneficial if people use it efficiently. In the case of the DWH, this means that old data sources should be able to be replaced by the DWH as far as possible, and that the DWH should be the central hub for data access and data evaluation for most users. In addition, responsibilities, roles, processes and guidelines for dealing with the new IT landscape had to be worked out so that the customer would actually benefit. 

True to the motto “Support the Mind Change”, our concept therefore includes a component that explicitly aims to pick up future users of the solutions to be developed in good time, to record their requirements at an early stage and to incorporate them into the design of the solution. Finally, we lay great emphasis towards offering comprehensive application documentation, providing customized training materials, and conducting regular question and answer sessions. 

The overall concept was then presented to the customer. In the meantime, various approaches to the further procedure were explained and presented to the customer as a basis for decision-making. The customer ultimately decided to follow our advice and implement the concept in an agile manner in a more extensive follow-up project – in order to get the research and development for Data Science ready for take-off.