Process Mining vs. Business Process Discovery
As a boutique media company, we like to follow in the footsteps of others who have already figured things out. One of those companies is Skift, a media firm that covers one of our favorite topics outside of technology investing, travel. Slogans like “doing less with more is the new doing more with less,” or “going slow to go fast is the new scaling up,” are right up our alley considering we’ve hired a bunch of underpaid MBAs to publish insightful content that we give away to everyone for free.
Another goal they have over at Skift is to avoid industry buzzwords so that their content remains timeless. That’s especially important in the fast-moving world of disruptive technologies. In past articles, we’ve looked at how grandiose sounding names can mask simple underlying technologies. Edge computing? That’s just moving cloud-based applications closer to the devices that use them. Cloud computing? That’s just running applications on large servers which everyone shares using workstations. These are all ideas that have been around for decades. As they like to say on the banana pancake trail, “same same, but different.”
We recently came across another term – process mining – that seems to have an infinite number of variations. Before we crawl down that rabbit hole, let’s start with some basic definitions.
A business process is a set of repeatable steps that occur within an organization that you might be able to map using a workflow diagram. Here’s an example of a simple business process:
This simple business process shows what happens when a business department has a new job opening and needs Gwenyth in Human Resources to put together a wildly inaccurate job posting which can then be used to source candidates once it’s gone back and forth between both departments about 76 times. Defining such a process falls under the broad umbrella of something called “business process management” or BPM.
What’s Business Process Management?
A business process analyst (BPA) whose job is to perform business process modeling would have built the above workflow diagram to capture the business process as part of a business process management initiative. The reason to do so is quite simple. If you’re a department head tasked with managing a number of teams who perform complex tasks that you wish to make more efficient, you’d want to map all the business processes first. A prerequisite to business process automation (BPA) or robotic process automation (RPA) is defining repeatable business processes. We previously waxed poetic about the term “robotic process automation” which is simply just “software automation” with some pictures of robots splashed on the marketing material and the use of some machine learning algorithms. Here’s an excerpt from our prior article which asked, “Is Robotic Process Automation (RPA) Just Marketing Speak?”
You remember back when you called your banking customer support line and some guy named Darrell with an Indian accent started reading through a script? Now, when you call your bank, it’s a friendly lady name Maria with lovely manners who won’t tell you she’s based out of the Philippines.
Maria still uses that same script Darrell did, except now companies are starting to think – if all these functions are so scripted, why don’t we just automate them with software? Then, we can get a whole bunch of robot pictures to splash all over our homepage and re-brand what has already been going on now for decades as “enterprise process robotics” or “robotic process automation” or any other combination of the words “enterprise/process/robotics/robots?”
It all boils down to making existing business processes more efficient and replacing increasingly sophisticated human tasks with automation when possible. That’s largely what business process management is all about.
Why Process Mining?
Harvard Business Review – a publication that’s, unfortunately, becoming increasingly political by the day – published an article about process mining earlier this year, written by two individuals who have been involved with the field for four decades now. According to the experts, process mining solves a few fundamental challenges associated with business process management. These are:
- Companies tend to spend too little time or too much time analyzing “as is” business processes
- There is a lack of connections between business processes and an organization’s enterprise information systems.
Starting with the first bullet point, we’d argue that for most companies, if you can’t figure out the optimal time to spend analyzing an existing business process, hire better BPAs. The second bullet point describes the inability to capture the “interoperability’ of processes and information systems. Fair enough. Organizations are incredibly complex entities, and one department might interact with hundreds of internal systems. Enter process mining and a German company called Celonis.
Celonis and Process Mining
Founded in 2011, German startup Celonis has taken in $367.5 million in funding with their latest round – a $290 million Series C – closing just this past month and placing the company on the CB Insights Unicorn List with a present-day valuation of $2.5 billion. The firm appears to be quite profitable, having been cash flow positive every year since it was founded with a customer list that includes names like Siemens, L’Oréal, Uber, Citi, Airbus, 3M, Cisco, and Vodafone. They’re the world leader in process mining according to Gartner, and their tools are used to make processes more efficient across all areas of an organization – finance, supply chain, IT, sales, and customer service. The Celonis definition of process mining offers some hints as to how their solution works:
Process mining is an analytical discipline for discovering, monitoring, and improving real processes (i.e., not assumed processes) by extracting knowledge from event logs readily available in today’s information systems.
Here we can see that they’re analyzing application event logs to extract insights. (Event logs contain detailed information about everything that happens within an enterprise application, as it happens.) Putting together a workflow diagram captures an “assumed process,” while analyzing an event log shows a “real process” by examining how it works in real life. Says the company, “even the best plans have exceptions, and over time these exceptions can become the rule.” If you’ve served hard time at any large corporation, this notion of looking at what “actually happens” vs. what humans “say happens” is starting to ring a bell. It all starts with something called “Process Discovery” which involves unearthing inefficiencies.
Identifying where all the muck-ups are happening, and then fixing the inefficiencies, is just the first step. What’s more important is ensuring that inefficiencies don’t creep back into your processes. That’s where some robotic overlords can come in handy.
The Celonis Action Engine is a new kind of process excellence tool—an AI-powered process assistant that constantly analyzes data, and then sends signals to everyone involved to let them know that they need to take action to get a process back on track.
Why stop there? Celonis now wants to take process mining to the next level by looking at something called “task mining.”
What Would You Say, You Do Here?
Many readers are familiar with that scene in the cult classic, Office Space, where the Bobs (a couple of efficiency experts) are interviewing an employee and it’s pretty obvious his job doesn’t actually entail doing anything of value. What your employees say they do, and what they actually do, are widely disconnected. This is where task mining comes into play.
“Task mining is not about spying on your employees to find out who’s spending the most time on Facebook,” says the company, though it’s pretty clear that you’re not going to be spending much time on Facebook when this tool gets deployed on your desktop. It’s all about automatically detecting the systems users are interacting with – user interaction data as it’s called.
Above we can see how the tool records your interactions with various applications, uses optical character recognition to read what’s on your screen, and then feeds this to AI algorithms which can then “understand” what it is you say, you do here. That information can then be attached to business processes in order to identify
slackers ways to create efficiencies. It’s also a great way to figure out what mundane repetitive tasks are being performed by humans that might be replaced with some good old-fashioned robotic process automation.
Process Mining vs. Business Process Discovery
Now that we understand what process mining is, it’s important that we understand what’s coming next so that all you CTOs out there can leapfrog today’s trend and show the board what a “thought leader” you are. For this, we’ll turn to another large German startup called Signavio that provides tools for business process management and has taken in just under $230 million to that aim. Their homepage is rife with buzzwords, but they do much the same thing that Celonis does – process mining. In an article on their website that appears to have been published last year, “Signavio Co-founder and CEO Dr. Gero Decker argued that stand-alone Process Mining will be set to disappear in 2019, or at the very least, be replaced and upgraded to something slightly more actionable: Business Process Discovery.” In other words, we’re now moving from “process mining” to “business process discovery.”
And that’s what we used to call hiring a business analyst to map business processes – “business process discovery.” Even more than that, Signavio suggests that we move to “automated business process discovery (ABPD).” That just so happens to be what Gartner suggested back in 2008 in a paper about how “automated business process discovery” improves business process management outcomes.
When it comes to nomenclature, we’re now back to where we started over a decade ago.
Process mining means that we don’t just hire a business analyst to create workflow diagrams that define how processes ought to work. Instead, we use tools that analyze event logs from enterprise applications so that we can see how processes actually work in the real world. Then, we can easily identify the inefficiencies that exist and fix them. Once that’s sorted, we can then have some machine algorithms make sure that other inefficiencies don’t surface themselves over time since processes are always changing. At a more tactical level, we can use something called task mining to find out what employees spend their time doing. We can then automate repetitive tasks using robotic process automation tools and create jobs where everyone adds value to an organization. It’s like removing jobs that are dull, dirty, and dangerous. Do it right and everyone wins.
We hope you enjoyed reading this article as much as we enjoyed writing it. Did you know you can hire our team of MBAs to write about your exciting disruptive tech story? Check it out.