At TM Forum Live! (in Nice in May), Orange Caribbean and sOftbridge technology will deliver a case study on process mining. In advance of the show and their presentation, Patrick Charra, CIO, Orange Caribbean, gives an introduction to the topic.

Tracking in hunting is the science and art of observing animal tracks and other signs, with the goal of gaining understanding of the landscape and the quarry. Another goal of tracking is a deeper understanding of the systems and patterns that make up the environment.

The art of tracking is the first implementation of science, practiced by hunter-gatherers hundreds of thousands of years ago. We always have to remember where we have come from, and returning to our ancestral practices may help us to understand the digital world.

Business processes running over information systems are like animals walking on the ground. They leave numerical spoors in the applications they cross. The skilled tracker is able to discern these clues and recreate what happened on the landscape, to retrieve the animal and make predictions about the next steps on the quarry’s path.

In our digital world, these clues can take various forms: trace files, database records, log files, event logs, etc. By observing our numerical landscape, we can find these clues, collect and analyze them to recreate the business processes as they are in the real life and not as they were supposed to be, as designed by engineers.

It has taken years for us to be able to apply the tracking techniques used by our ancestor cavemen in their day-to-day life to our numerical world. Over time, we have become able to better control, understand and manage our digital world.

Big data technologies have brought us the ability to collect and analyze large volumes of data. At the same time, artificial intelligence (AI) has provided the software, learning and deductive capabilities. The combination of both gives hunters of modern times the tools to observe, analyze and understand what happens in our digital world.

Process mining is one modern version of the ancestral art of tracking. It is a particular field of data mining that looks at analyzing the data produced by information systems with the objective to discover, control and optimize business processes. Professor Will Van Der Aalst and his team at Eindhoven University carried out the most important scientific advances in this field for the last 10 years. He wrote the books Process Mining: Discovery, conformance and enhancement of business processes and Process Mining: Data science in action.

Data provided by information system operations are used as a data source. They are collected and analyzed in real time to extract information used to build the process instances (also known as cases). Thanks to process mining algorithms and machine learning techniques, cases are analyzed and characterized to model business processes (process). It is then possible to check the conformance of each new process instance (case) with the nominal process (process).

Your business processes and the way your information system works are then revealed to you: places where time is wasted and places where errors occur in a particular operational context. It is also possible to detect failures or bottlenecks and optimize business processes.

Fig 1: Process mining principles

There are many advantages to using process mining to control and optimize business processes.

First of all, it is a non-intrusive approach which reveals the reality of business processes. Business processes are extracted from real data provided by your operational environment and reflect the processes as they are, not as they are supposed to be.

Next, you don’t have to describe what you want to monitor. The system discovers business processes by itself, thanks to learning machine techniques. This also gives solutions a remarkable robustness when it comes to changes. When a new release of an application is deployed, any changes are automatically detected and business processes are updated to reflect the new way of operating.

Finally, process mining provides a real-time, up-to-date and unified view of both business processes and information system operations, thus bridging the gap between IT teams and end users.

At Orange Caribbean we have used the sOftbridge technology solution, SME (Smart Monitoring Engine), to manage all our touchpoints with our customers.

This includes all our interaction channels:

  • Point of Sales (order capture application)
  • Web (online shop and web support)
  • Mobile app (Orange & Moi)
  • Call center (CRM)
  • Social networks (Dimelo)
  • Mobile self-care (USSD, IVR)

… and back offices applications are observed by sOftbridge.

Thanks to big data technologies embedded in the sOftbridge solution, we are able to analyze very large amounts of data every day to discover and control every process instance initiated on any interaction channel.

We detect non-conformant process instances in real time and check what is wrong without wasting hours trying to locate the root cause. Conversely, we know in real-time the impact of any technical failure on the customer.

Regarding operational KPIs, we have decreased the failure rate of our order-to-cash process from 4.87 percent to less than 0.2 percent. We’ve reduced our end-to-end time to deliver for a new subscription from 33 minutes to less than 9 minutes. The number of customer complaints managed by our call centers has been halved.

However, discovering and controlling business processes is not the ultimate objective of our quest. When you put all the process instances initiated on all the interactions channels together with relevant information on how the customer demands were met, you create a clear and accurate picture of the customer journey. There is no doubt that being able to provide an up-to-date view of the customer journey will be a key challenge for IT teams in the next few years. Achieving this will be a powerful weapon to help marketing teams to succeed in increasingly competitive markets.

This is our next prey and we are on its tracks …


This article first appeared in TMForum