AkzoNobel is the world’s largest paint and coatings company and a major producer of specialty chemicals. Like many other large companies, AkzoNobel has done its best to document its processes with hopes of maintaining and improving quality and boosting efficiency.
Recently, though, the company discovered that it could better understand its processes—and do so with less effort—by using newly developed business process mining technology. Specifically, the AkzoNobel Decorative Paints (ANDP) unit wanted to better understand their procurement processes, which are standardized in SAP. Over the years, the company’s local entities, scattered across 80 countries, had adopted variations on the standard company process to better meet local needs.
In an attempt to understand and manage that variation, the Decorative Paints unit initiated a “value extraction” program focused specifically on determining whether there were deviations from the standard company processes that are not desirable from a compliance or control standpoint—potentially putting the company at risk from a regulatory standpoint, or introducing inefficiencies. The effort also sought to find out whether corporate best practices could be enhanced by lessons learned locally within the local entities of the Decorative Paints unit.
With the help of CapGemini, a consulting firm, ANDP adopted Disco business process mining software from a company called Fluxicon, one of the pioneers in the new field. According to Fluxicon, which published a case study of the project with ANDP, the application first extracted thousands of purchasing events from the SAP system. This data was then analyzed and visualized by CapGemini using Disco and the resulting visualizations helped pinpoint non-standard practices, which were then further investigated to determine their origins and results. The team accomplished this work in a fraction of the time and with a fraction of the labor needed for traditional manual processes according to Fluxicon co-founder, Anne Rozinat.
What’s more, notes Rozinat, who was among the early researchers in the field at Eindhoven University of Technology, unlike slow, manual methods, the results are also provided much closer to real time.
Tools to Spotlight ‘Hidden’ Business Process Data
Fluxicon is not alone in taking process mining into commercial practice. Among the other vendors in the process mining space are Futura Technology, Fujitsu, Ltd. (with its Interstage Integration Middleware Suite), the Pallas Athena BPMone software suite, Software AG (as part of the Process Intelligence Solution), QPR Software Plc, Celonis GMBH, and Perceptive Software.
To help promote process mining, Perceptive and Fluxicon sponsor an annual competition, called the Business Process Intelligence Challenge. This year’s challenge was won by CKM Advisors, a New York-based consulting firm. In a presentation posted by Fluxicon, Lalit Wangikar, a partner at CKM, explains that process mining is different in many ways from traditional methods of analyzing business practices. For instance, he notes, most of the traditional methods of analysis depend on existing process documents and interviews—or data samples rather than full data sets. “What that ends up leaving on the side is the exception handling that goes on and all the hidden ‘factories,’… that exist in an organization,” he said. Those “factories”—the undocumented exceptions to standard processes—can represent a substantial portion of business activity.
“What we find is that through traditional means you get a great understanding of the “average process”, so maybe 60 percent to 70 percent of work items go through the average process, but the challenge is the exceptions or the things that do not follow that average path,” notes Wangikar.
Inevitably, that’s where you will get the most improvement opportunities, he says—where there are a large number of exceptions.
“Process mining, as a new tool to look at event level data for a large number of work items, gets you to a level of detail that you can start covering these exceptions in a very meaningful manner, and you can show it very tangibly to a business manager or a process owner. That’s the benefit of using process mining,” Wangikar adds.
Rozinat, who is from Germany, says the technology was first developed and fostered in the Netherlands, starting about a decade ago. “When researchers at Eindhoven University said they thought they could reconstruct business processes from a database or log file; everyone thought they were crazy,” she says. The first successes of the methods were then dismissed as a mere academic exercise. However, Rozinat says businesses are now getting real value from process mining.
Business process mining “can really help people define and understand process problems, which makes it different from other analytic approaches,” she says. For example, data mining and traditional analytics might help show what products are being bought and by whom – but they won’t reveal much about the buying process, Rozinat says.
“What I am seeing and what people tell me is that in business there is often a lack of transparency into processes. People document them and spend a lot of time on that but the process maps that result are often not the same as the actual process,” she says. Not only are actual processes more complex, they are often in flux, leaving even the best traditional documentation efforts far behind.
“Process mining bridges that gap and applies analytics to help people visualize,” she explains.
Rozinat acknowledges that all processes are not created equal when it comes to the usefulness of business process mining. For instance, she notes, service processes generally seem to benefit most from the analytical approach, in part because they often have less documentation associated with them, and do not have as many controls as a factory floor process. That fact also means they can present substantial optimization opportunities. Rozinat says government administration practices are also popular subjects for process mining.
On the horizon, according to Rozinat, is process mining that can be applied to healthcare. “There is a lot of research right now in that area; every patient is unique so there are very diverse and complex processes.” And, in the meantime, Rozinat sees many opportunities for process mining to demonstrate its business value.
Alan R. Earls is a business and technology writer based near Boston.