Follow Datanami:
December 13, 2016

Business Process Mining Promises Big Data Payoff


As companies grow, behind-the-scenes business processes get more complex. And as the complexity grows, it takes a toll on efficiency. With the advent of big data analytics, companies are finding that techniques like machine learning can help identify inefficient processes and reclaim millions in lost productivity.

Business process modeling (BPM) is not a new discipline. But thanks to rapidly emerging analytic technologies, a new class of vendors is emerging to give companies even more power to track and measure how well business processes are working at a very granular level.

One of the new process miners staking a claim in enterprise is Celonis. The company was born in Germany 5.5 years ago with the idea that companies can weed out inefficient processes fairly easily using the power of software. By mapping the actual flow of work among all the different back-office applications and comparing it to ideal workflow scenarios, Celonis can identify bottlenecks and other business pain points that leak profits.

“Honestly, what we’re doing is removing the need to pay a Deloitte or McKinsey to come in and do all this discovery and find the issue,” says Celonis Chief Marketing Officer Sam Werner. “We can replace the expensive consultants that get paid today to go do this work, which by the way can be very disruptive for your organization.”

Process mining starts with data discovery. Its product, called Proactive Insights (or PI), Celonis scans mainstream ERP systems from the likes of SAP, Oracle, and Microsoft, identifies the business processes, and then constructs a visual representation of how the process works. This gives architects a low-level view of how many different places each individual purchase order waits for approval, for example.

If you’ve ever seen a BPM tool, you know that visualizations of business processes can get quite messy. The so-called “process spaghetti” demonstrates that, while there typically is a standard way to do things, that variations inevitably creep into the process. Sometimes these variations are required, such as for compliance purposes. But other times, the variations are unnecessary shortcuts, or even instances of outright fraud.

But tracking the flow of every business item is a big undertaking. That’s where big data analytics comes in. After mapping the actual workflow, Celonis applies data science techniques, like statistical analysis and machine learning, to tease the inefficiencies out of the resulting data.

Celonis mapped the flow of nearly 280,000 purchase orders for one company, and this is how it looked.

As Werner explains, it’s all about business process optimization and transformation. “Companies are not delivering on their business commitment to customers, and there’s very avoidable things that are contributing to that,” he tells Datanami. “Maybe it’s slow handoffs. Maybe it’s unnecessary approvals. Maybe it’s selecting the wrong kind of shipping based on the required delivery day. There are lots of different contributing factors.”

The company, which recently opened a new office in the United States, last week announced four new PI modules, including PI Machine Learning, PI Conformance, PI Social, and PI Companion, to help customers eradicate inefficient processes, which Werner says eats up 10% to 20% of the average company’s profits each year.

Companies have tried to automate their processes with rule-based triggers for many years. But the power of machine learning promises to take that automation to a whole new level, says Werner.

“I’m not quite sure how you would” do this without machine learning, Werner says. “Most people set static thresholds for manufacturing that say, when I go below this threshold, But you’re not able to adjust it dynamically based on performance…You don’t learn from actual performance and adjust. To me, that’s the big difference, is learning in real time, in your day to day operations.”

Another company tackling the big data opportunity in business process optimization is data modeling firm erwin, which just acquired BPM modeler Casewise. According to Martin Ownen, vice president of product strategy for erwin, big data is set to have a big impact on the BPM field.

“Big data enables an organization to extract value out of the vast and diverse quantities of data in its environment,” he tells Datanami. “When combining process modeling with data, we can understand the data in the context of the business processes, where it is used, by whom and how. Analyzing data without this perspective makes meaningful comprehension and analytics next to impossible, as the data cannot be quantified or prioritized from a business perspective.”

The company is in the process of integrating the erwin data modeling tools with the BPM tools from Casewise. “The two products go hand in hand,” Ownen says. “From a process perspective, you will be able to see what data is used in the process and architecture models, which gives more insight to the Casewise user. From the erwin and data perspective, users can see the context of the data and the impact of all the places it is used in processes across the enterprise.”

As the ramifications of big data analytics continues to ripple through our lives, it’s clear that no industry will be kept out of view—not even the mundane business processes that are the back bone of human interaction with large enterprises. If there’s a way to eek greater profits by driving up business process efficiency, then big data analytics will undoubtedly find a way.

Related Items:

Making ERP Better with Big Data

How Deloitte Transforms Clients into Insight-Driven Organizations

Hiring with Hadoop: How Big Data is Helping HR