How Deutsche Telekom scaled automation quickly through RPA
Learn about Deutsche Telekom’s automation journey and how it applied RPA to achieve an enhanced, centralized automation governanceAdd bookmark
Global integrated telecommunications provider Deutsche Telekom manages one of Europe’s largest robotic process automation (RPA) implementations and has automated more than 450 processes managed by 3,000 unattended bots since 2015.
Deutsche Telekom’s service division, handling 100 million customer requests each year, is attempting to implement an initiative that will see it gradually transition from front-end to back-end automation and replace bots with APIs. The telecommunications company is undertaking works to further develop automation capabilities, despite achieving savings of more than €93m (US$113m) last year through RPA.
Deutsche Telekom endeavored to tackle pain points in manual customer service processes in 2015 when it began its journey toward digital transformation, explains Marco Einacker, vice president for service IT at the telecoms giant. However, RPA was one of the few technologies that it was possible to implement, as the business was operating with underlying legacy IT systems which did not provide APIs.
The business experienced swift and significant results over the following years and became one of the biggest RPA users in Europe as it scaled up RPA automation in different lines of business. By 2019, it had automated more than 450 processes, managed by 3,000 unattended bots, achieving savings of more than €93mn in 2019.
Einacker said: “RPA is a good thing and it has positive effects on our organization, bottom line and on customers and employee satisfaction.”
The downside of fast solutions
Despite initial and significant successes, Deutsche Telekom ended up with an automation system difficult to manage. With seven different RPA platforms and seven different libraries to maintain, the business began to experience issues at scale. The complex nature of this automation setup meant that a single change in the main CRM system led to change requests on four different RPA platforms.
“Over the years we automated more and more complex processes so code became harder to maintain,” explained Einacker. “We feared the risk of technical debt as we had no back-end automation.”
With Deutsche Telekom’s business knowledge strongly linked with RPA code, it began to experience rapid growth in limitations on an organizational level. Developers had become the business process experts simply because they had built the processes in code. Due to the nature of RPA technology, it could not be easily combined with different tasks – such as manual task – or RPA technologies from different vendors, Deutsche Telekom began to experience major problems as it looked to implement end-to-end processes.
In 2018, Deutsche Telekom undertook a new strategy in an attempt to alleviate the challenges it had been experiencing. The telecoms provider looked to reinvent automation, with the goal of separating processes from the technical layer and moving to back-end automation.
Through the consolidation of the seven RPA platforms into three, selected due to their high scalability and functionality, the business attained a new centralized governance. It allowed for collective decisions on whether future solutions should be bot-driven or become a core IT solution.
Deutsche Telekom developed a new platform, OREO, to separate the process and technical layer. Camunda’s workflow engine was implemented to handle all business processes in business process model and notation (BPMN) and decision model and notation (DMN), and to orchestrate RPA and user tasks across end-to-end processes.
Business experts were able to cut out unnecessary tasks and refine individual steps from old RPA-driven processes, reinventing them. Following this, developers were able to automate each step of the process.
Camunda provides a common language for business experts and developers in the form of BPMN. This globally recognized graphical language can make things easier for developers before they even start programming, resulting in significant development time savings due to a reduced amount of necessary code to be written. By slicing up each automated step, developers can decide if a RPA bot or a human task is the best solution for any step in the process.
“BPMN models are process optimization engines for us, we gain transparency into how complex a process is, and we get ideas on how to simplify it,” notes Einacker. “With this end-to-end view andthis increased transparency, we are able to make process improvements.”
End-to-end visibility and orchestration
The next step in Deutsche Telekom’s journey is to eliminate the need for bots entirely and build APIs, now it has achieved the separation of the RPA and technical layer, moving from front-end to back-end automation.
If you are interested in a deeper dive into Deutsche Telekom’s journey, you can watch Marco Einacker’s CamundaCon LIVE presentation on-demand.
This article was originally published as How Deutsche Telekom scaled automation fast – using RPA and Camunda.