To understand Process Mining, imagine the following situation:
Ben’s mother calls his cell phone and asks him to go to the grocery store (which is two blocks from his house) and buy:
- 3 dozen eggs;
- 6 lemons;
- 1 pork shank weighing 8kg;
- 2 packages of 1kg of fried chicken;
- 2 boxes of powdered soap;
- 3 jars of canned olives (note: large jars);
- 1 package with 5kg of rice;
- 1 bundle with 12 cartons of 1l of milk.
Also, this unusually big shopping list must be purchased and brought back within 20 minutes.
After 22 minutes have passed, Ben’s mother is already home and he has not yet returned from the shop with the groceries.
After an hour of waiting, he arrives completely exhausted carrying all the items on the list. His mother questions him:
– What took you so long? It was just a few “things”!
To which he promptly replies:
– When you called me I was sweeping the garden. I had to get some water, change my clothes and close up the house. After that, I walked to the grocery store. Today is Saturday and yesterday was payday. It means that the market was more crowded than usual, with long rows to the cashiers. I don’t usually shop, so I had to look more for the items. Because of the movement, it was difficult to get help. Finally, there were no cars available for delivery in less than 2 hours (for this example, disregard that he could have called an Uber, or cab, or asked his parents to pick him up).
Applying Process Mining to the example
At first glance, the process is clear and very simple:
Process: Receive the call; Write down what should be bought; Go to the grocery store; Complete the task in exactly 20 minutes.
Execution: Receive the call; Drink water; Change clothes; Close the house; Walk to the market; Market fuller than normal; Inexperience with shopping; No cars available for delivery in 2 hours; Number of items to be brought; Maximum time for execution 20 minutes.
This completely ludic example shows that, in any situation, processes that look simple in the eyes of the planner can become complex at the moment of execution.
So, even if there were many other things to be done (which possibly delayed the process), we can effectively question:
- How much time was spent on each of the tasks described by Ben?
- How much effort went into the overall execution?
- Where was the biggest bottleneck in this operation?
To answer these questions we resort to Process Mining.
What is?
Let’s leave aside the example of marketplace shopping and transport ourselves to business environments.
By doing so, our examples grow exponentially and we begin to truly understand the importance of this tool.
To mine, from its meaning, is: to extract or exploit ores or valuable metals from mines.
Therefore, mining a process means finding valuable information in its structures and monitoring it in order to continuously optimize it. It involves eliminating bottlenecks and dispensable or non-value-adding tasks as well.
So, to use another analogy, the “mine” is any process that generates logs. Consequently, the ores (valuable insights) will be mined in a raw form.
Bringing it into the business environment, they are read into schemas, and cubes – running on a server. These servers can be OLAP working in parallel with an OLTP. In summary, this action ensures that there is no performance drop in the production environment.
Therefore, the “miner” is the system used by the organization. And then, finally, the product obtained is the meaningful data of this process in its entirety.
How does Process Mining work?
Often, when companies start their activities, and for a considerable period of time, their processes take place in an informal and manual way.
Thus, there is no specific order for things to happen. Therefore, they work between e-mails, spreadsheets, or chat conversations. Even potentially important decisions are made over the coffee break.
This scenario represents what we call AS-IS, which means exactly that: as it is.
But at a certain point, it becomes necessary to look inward and identify how things happen, and how they should happen.
Then, we can formalize, design, and execute processes in a more organized way, often using tools available in the market.
Benefits of Process Mining Tools
These tools, besides revealing the operation setup, will measure event logs to extract related information, such as execution times and other pertinent information.
Thus, this information can generate insights for decision-making, improve the efficiency of policies, or even create new and more updated ones. In addition to that, it ensures that decision-making is supported by correct data, bringing greater reliability to the paths to be taken.
It even creates a reliable history for audit purposes. This part of the process is called TO-BE: what will be, or how it should be.
All this can be done with a large mass of data stored in logs, which may contain several years’ worth of information, depending on the company’s age.
Similarly, in a relatively young company, but one that does a lot of daily operations, the volume of these operations can be considered, even if in a shorter time range.
Thus, if a company measures its processes in hours, with a normal working day, there will be around 8 hours of stored data to be mined and analyzed per day.
Through metrics that use counts or statistical measures, for example, data can reach very specific granularities.
This opens up a very comprehensive range of information and generates a basis for decision-making or adjustments in zero priority.
Process mining pillars
Let’s delve a little deeper into the function, types, and/or pillars of process mining. As they can be cited in different sources, we will draw some existing parallels in the literature on the topic:
- Process discovery: a technique in which an algorithm discovers a process’ model through logs;
- Conformity checking: comparing the realities of execution with the process modeling;
- Process Reengineering: reformulating a process based on the data obtained by mining;
- Operational Support: analysis, identification, and possible suggestions for process improvements in real-time.
The observations obtained by mining are expressed in graphs, tables, and flowcharts, among other kinds of visualizations.
Difference between Process Mining and BI
If you have a more data-oriented outlook, you can check what has been said so far and ask yourself: Is process mining the same as BI (Business Intelligence)?
The answer is: No.
BI aims through data to point out that there is a problem, a bottleneck, something non-compliant.
However, mining goes deep into this data and shows where the problems were the root cause of rework or the inefficiency of a process.
Going back to the example at the beginning, the BI would be the child that contains the history of the reasons for the delay. Consequently, mining would analyze each one of them, to actually understand if everything that was pointed out impacted the delay.
Furthermore, it weights the points to consider their influence on the whole, mainly, on the final result of the process.
Conclusion
In conclusion, using Process Mining is an important ally to guide your company on solid paths, since it’s a living organism and must be constantly evolving without leaving behind the lessons learned in the past.
In the same way, recent experiences show how the path is being followed, and also if what is being done is still effective or if it’s necessary to implement more innovations to achieve set goals.
And what about your company, do you already use Process Mining? Tell us in the comments about your experience with the tool!