Back to Hub

Achieving a Successful Robotic Process Automation Implementation: A Case Study of Vodafone and Celonis

06/07/2018 By

Procurement organizations are beginning to catch on to robotic process automation (RPA) and the potential benefits that the technology can bring, which include cost reduction and more time for strategic, value-add activities. In fact, more than a third of procurement organizations are already piloting the technology, according to research from The Hackett Group.

This was certainly the case for British multinational telecom corporation Vodafone, for which adoption of RPA quickly yielded results. Before adopting RPA, Vodafone completed 73% of its purchase orders correctly the first time around. But within six months of implementing the process mining and RPA software of Celonis, a provider that specializes in the technology, Vodafone had achieved a perfect PO rate of 85%.

This improvement led to an initial reduction in operational purchasing costs by 11% and a reduced time to market by 20%, all in the first six months. Today, Vodafone boasts a perfect PO rate of 92%.

A Brief Background 

Since its founding in 2011, Munich-based Celonis has undergone truly impressive growth. The solution provider opened its first international office in 2012, landed a number of clients among the Fortune 500 by 2014, reached a reseller agreement with SAP in 2015 and received $27.5 million in funding in 2016. In Celonis’s first four years, its revenue increased by 4,000%. And this year, Celonis is also on Spend Matters’ list of 50 Providers to Watch.

Celonis’s focus is on process mining, a type of data mining that identifies patterns in event log data to optimize process efficiency. Alexander Rinke, co-founder and co-CEO of Celonis, says that process mining offers “a much more controlled approach to RPA.”

“The technology uses the data accumulated within an organization’s IT systems to understand the current activities taking place by automatically reconstructing the way the organization is functioning, and identifying the opportunities for improvement,” Rinke explains.

And Vodafone, of course, is the household-name telecom company with some 470 million mobile customers, 14 million broadband customers, mobile operations in 26 countries and fixed broadband operations in 17 countries. Founded in 1991, Vodafone operates in Europe, the Middle East, Africa and Asia Pacific.

Choosing an RPA Provider

“Although Vodafone is in a leading position, it is still affected by telecommunications’ inherently short technology and innovation life cycles,” says Israel Exposito, Vodafone’s global process lead of big data. “RPA became attractive not just because of the opportunity to reduce costs but also because it liberates our employees from boring and repetitive tasks, giving them back time to focus on more interesting activities that require higher competencies. Nobody wants to spend years and years of study to wind up in a job doing copy-paste over and over again.”

Another challenge, Exposito notes, had been to make sense of Vodafone’s enormous amount of data. The company manages more than 700,000 purchase orders, 5 million invoices and 40 million assets annually, all together making up more than 15 terabytes of data.

“We discovered that traditional analytics tools could not uncover hidden inefficiencies or provide full transparency into our data, and we needed unbiased analysis to solve problems we did not even know we had, and to identify the most impactful opportunities for improvement,” he says. Vodafone needed to close its information gap and connect KPIs to root causes.

This is where Celonis came in. Vodafone went with the “classic RPA project management approach,” says Jesús Vita-Martínez, project manager of RPA at Vodafone. “We start with an assessment of the process within scope, and we take a call on moving to action by balancing cost and benefits. And finally we move to execution by developing the bots.”

Vodafone implemented Celonis in 2013, and since then, Vodafone has been able to look more deeply into its process variations, non-compliances and workarounds, as well as better compare results with goals, Exposito says. The combination of process mining and RPA technologies also helped Vodafone surpass its goal of a perfect PO rate of 80%, eventually achieving 92%.

However, Vodafone is not risking complacency. “RPA is the first step in a long-term strategy toward cognitive and artificial intelligence being used across Vodafone’s support services,” Exposito says. Currently, Vodafone is working with Celonis to combing process mining and RPA for a proactive alerting and remedy system.

Challenges of RPA Implementation

For organizations interested in adopting RPA or process mining, it is crucial to first have full transparency into business processes and know which processes would actually benefit from automation, Rinke warns.

“The risk of failure comes with a high price tag, and for every $1 you spend on RPA technology you can expect to spend $10 on consulting and implementation costs,” Rinke explains. “If a process is already flawed, RPA will only make a bad process faster. It’s essential that businesses take a transparent approach to RPA, ensuring that automated processes are first optimized to be the most efficient.”

Before embarking on an RPA initiative, Rinke advises, companies should know the answers to the following questions: Which processes are mature enough for automation? Which processes require the most manual rework? Which processes, if automated, would provide the biggest impact to the business?

Fortunately for Vodafone, the implementation process was smooth, and the challenges that came up were to be expected.

“As RPA is a relatively new science, the biggest issue [had been] creating an ecosystem — human resources, company policies, mindset, etc. — from scratch, which is necessary for an initiative to be adopted by [a large company] like Vodafone,” Vita-Martínez says.

It is not rare for RPA implementations to fail. Ernst and Young has estimated that 30%–50% of initial RPA implementations are not successful, but whether and how preventable those outcomes were is a different matter.

Beyond thoroughly understanding their processes, organizations interested in RPA should also start small and simplify those processes as much as possible. “Automating a simple process is exponentially more likely to be successful than automating a complex process,” Rinke says.

Rinke likens the decision to implement RPA to the implantation of a medical device. “You wouldn’t want [the doctor] to start cutting and inserting without first doing a diagnostic scan, and processes should be treated the same.”