Listen to this article
When Dutch financial services group APG tried to understand better why clients rang up its call centre, they noticed a small group taking up a lot of time: a high number of divorcees had been calling in the run-up to their legal separation, asking an expert to explain the intricacies of pension entitlements.
This was unwelcome news for APG: each call cost an average of €13 and it already had a website which it thought was tailored to clients, so calls were unnecessary. “We were trying to find out: why do people need to call three or four times, in some cases even more?” says Rachid El Hasnaoui, a business intelligence specialist at APG.
The solution, he says, was to employ “process mining”, an emerging field in big data analytics that is helping some of the world’s biggest companies — including GM, Airbus, KPMG and UBS — to automate tasks and adapt to the digitisation of industry.
Early adopters say the concept will eventually be as common as terms such as Industry 4.0 and predictive maintenance. The idea is to gather all of a company’s internal decisions, from invoicing to logistics, and then visualise them as “digital footprints”. Process mining maps these images into flowcharts that can be filtered and reworked, offering an overall view of what steps are taken following any decision. It works in real time or by looking backwards.
“It’s a brand new science — five years ago nobody talked about process mining. These kind of capabilities did not exist,” says Israel Expósito-Peraza, who oversees an eight-member big data group at telecoms group Vodafone.
APG used process mining to aggregate data from their clients’ search terms, emails and phone calls, then sequenced them to create a “customer journey”. It could then detect patterns and personalise its website, making future calls from clients in a similar position unnecessary. It was one thing to gear the site toward the needs of a 25-year old or someone nearing retirement, but quite another to suit a client with a failing marriage.
“Before we could only see part of the picture, and we didn’t know the context,” says Mr El Hasnaoui. “We can [now] see the whole flow of the customer journey, the whole lifecycle . . . if the customer is getting married, getting divorced, we can even look at if the customer is getting worried. Those things have an impact on how we can be help the customer make the right choice.”
The concept of process mining emerged out of Dutch academia and has been put into practice by a few start-ups, including Celonis, a Munich-based software company founded in 2011 by three graduates of the Technical University of Munich. The trio now oversee 250 employees, up from 70 last December, with offices in Munich, the Netherlands and New York.
Bastian Nominacher, co-chief executive and co-founder, can sound a bit dreamy as he describes how process mining is “universally applicable” for more than 300,000 companies across the world. He estimates the global process mining market today at about €150m. But he reckons that by 2025 it will reach “at least a similar size as the business intelligence market”, which last year had revenue of €15bn.
Celonis says it has already acquired more than 350 clients across 15 industries — including Swedish robotics company ABB, railway operator Deutsche Bahn and consultancy Deloitte — and it has partnered with SAP, Europe’s largest software company, to offer process mining within a broader package of enterprise solutions. “There is no difference if you apply this is in Japan or the US,” Mr Nominacher says. “There are always the same challenges, always the same benefits.”
Lars Reinkemeyer, head of global process mining services at Siemens, says 4,000 employees are using Celonis to translate big data, which he describes as a lake of raw, undifferentiated files, into “smart data” — defined sequences for business processes, deliveries and customer orders. The visualisation allows Siemens to contrast best practices with loopholes and bottlenecks that can be improved.
As far as Mr Reinkemeyer knows, he is the only person anywhere to hold his job title. But he thinks that will not be true for long as more companies see the value in automating tasks. He describes process mining as giving Siemens X-ray images of its internal operations, making it easier to spot inefficiencies.
Users say discovering such anomalies usually requires an expensive team of consultants to diagnose them and recommend solutions. But in this case it is the consultants themselves whose jobs may be at risk, as teams of data engineers are now doing their own prognoses.
“We are actually being our own consultants,” says Vodafone’s Mr Expósito-Peraza. “A consultant can watch this and that and make recommendations. But once the report is done he says goodbye. Our team is finding the problem, finding solutions and driving the transformation.”
Vodafone, which has been using Celonis since 2015, says process mining has helped reduce costs per purchase order by 11 per cent and improved time to market by a fifth. Process mining has allowed Vodafone’s completion rate of “perfect deals” — internal orders done without any manual rework — to improve from 73 per cent to 92 per cent in two years.
For some clients, optimising internal processes is not just something that helps its business — it is the business. “We won’t be inventing a new iPhone tomorrow,” says Thomas Reichmann, manager of logistics at Schukat, a family-run supplier of electronics in Germany’s Mittelstand (its small and medium-sized enterprise sector). “As a wholesaler our processes are ultimately all that we have.”
The bulk of work at Celonis is building software to visualise and automate internal processes, supply chains and back-office management. Mr Nominacher says it is now moving into the more glamorous internet-of-things arena, enabling companies to track, say, a robot that picks up and distributes packages in a warehouse.
The group says it has been profitable since its first year, but to expand operations it has received funding from Accel, an early stage investor in Facebook and Dropbox, and 83North, an Israeli venture capital firm. Last year it raised $27.5m, a big amount for a Series A funding round. Mr Nominacher says he hopes Celonis will be Germany’s next “unicorn” and will have an IPO at a valuation above €1bn in 2020.
Get alerts on Big Data when a new story is published