Improve on-time delivery and delivery time with process mining

Process mining is a new analytical discipline for recognizing, monitoring and improving operational business processes in companies. Real processes are reconstructed in all their diversity using event logs from the operational ERP systems. The special feature is that no complex data transformations are required for these analyses: The common programs use standard connectors, are based on the unchangeable basic tables of the ERP systems and integrate master and transaction data on products, materials, customers and others.

Use of process mining in order processing for better adherence to delivery dates

These digital traces can be used to analyze process deviations. For the first time, process mining enables precise and comprehensive quantification:

• Which process steps have what share in the delivery time?
• Which sections of order processing are affected by deviations and loops and how often?
• Which process deviations cost how many percentage points in on-time delivery?
• Which products or customers are affected by the process deviations and to what extent?

Because the gain in delivery time and adherence to delivery dates can be precisely determined, investments can be validly justified using a business case for the first time. Companies can therefore make better economic decisions and make more targeted investments.

In addition, the respective best practice processes can be identified on the basis of performance. For example, if there is a product category that has a particularly high adherence to delivery dates compared to others, the differences in the process chain can be analyzed visually and quantitatively. Suboptimal process steps can now be adjusted iteratively.

Market overview of technology providers

As part of the SAP OpenEcoSystem and a certified partner of Celonis, Rothbaum uses Celonis Process Mining technology, which is why this article focuses specifically on Celonis and SAP. Beyond that, there are only a handful of serious providers of process mining technology. If you want to learn more about the technology itself and the providers, I recommend this article.

In the following, I would like to show you how you can set up process mining in your company and how you need to proceed in order to use this new technology to add value.

How to set up the process mining project correctly in order processing

Organizational framework conditions

The setup for a process mining project follows familiar patterns. In addition to the project manager, the specialist departments and, if necessary, external support, the IT department and the analysis team are particularly important. Process mining projects are very data-intensive at the beginning, as the digital traces of the processes are only found deep in the systems. You should therefore clarify the following questions at the very beginning and bring the necessary experts into your project team:

1. In which ERP and IT systems are the processes to be analyzed mapped?
2. By whom and where are the systems hosted? Are the systems administered internally or externally? How can the data transfer be organized?
3. Are individualized transactions or fields used in the process flow?

Basically, you have to decide whether the data model should be hosted internally (on-premises) or externally (off-premises, e.g. in a cloud). There are also two options for providing the data:

• Static: Deduction of data as a copy for a defined period of time
• Dynamic: Implementation of connectors that continuously and automatically copy the latest data from the system to a data warehouse.

To start with, I would advise you to choose a static data model that is hosted in-house, as the effort and complications are significantly lower – after all, process data often contains confidential information. Once you have gained experience with your company and have built up trust in the solution provider, dynamic and cloud-based solutions are preferable due to the better data quality and easier maintenance.

Relational data model for order processing in SAP

The following SAP tables are required for the data model of the order fulfillment process:

• VBAP: Sales document item data
• VBEP: Sales document schedule line data
• VBAK: Sales document header data

In addition, other tables can and should be linked to provide additional information, such as material master data (MARC) or customer master data (KNA1, KNB1).

As Celonis was developed specifically (but not exclusively) for SAP, the basic tables can be imported without further processing. No further manual data transformations are necessary. However, you will need an individually created “translation table” that defines the relevant events and names them in a generally understandable way.

The following figure shows an example of a relational database model:

Selection of the analysis and further customization

Before you can draw insights from your process mining model, you first need to create the appropriate analyses. Here, too, Celonis provides you with support and enables you to enter the world of Process Mining directly. Ready-made, pre-configured analyses can be installed from the Celonis application for many standard processes, including the order processing process. A total of over 400 ready-made analyses are available and new ones are constantly being added.

However, most companies use individualized SAP systems, which is why the preconfigured analyses in Celonis usually have to be developed further. These analyses give you an overview of your process and provide initial insights. Of course, you can adapt the analyses and add further elements for individual issues.

How to analyze order processing

As soon as the data is available and the necessary evaluations have been adjusted, you can move on to the most important part of the analysis: the deviation and root cause analysis.

The starting point for this is the Process Explorer. The Process Explorer determines the most common process flow within order processing. This usually corresponds to the desired target process. Based on this, you can gradually display the other process variants in order to visualize and understand all deviations from the target order processing process.

In the example process shown, 67% of the cases considered correspond to the target order processing process. By displaying the next two variants, the scope of consideration increases to 96% of all cases.

The next step is to recognize typical deviation patterns and focus on the cases that show patterns. Concentrate primarily on process steps that cause one or more loops or have a long lead time. You can use the Celonis lead time analysis for this.

Try to understand the causes and use the linked data sensibly. Delays in order processing may only occur with certain product groups or only certain regions or individual customers may be affected. When investigating a problematic product group, for example, you may discover that an important supplier is not fulfilling its delivery promise. You should carry out such a root cause analysis for every major deviation. Celonis enables you for the first time to determine the gain in on-time delivery due to the elimination of the problem in precise percentage points.

The right measures for better delivery performance

Once you have identified and investigated a number of deviations, you should next turn your attention to process improvement. First of all, you should cluster the existing deviations, as the same measures will often lead to success for similar deviations. This allows you to better quantify how much improvement you can achieve through a single measure (e.g. improving on-time delivery). The clusters help you to prioritize your measures.

Within the individual clusters, you should focus primarily on resolving bottlenecks and eliminating process loops. In the event of bottlenecks, you can, for example, expand your supplier network or optimize your stock of raw materials. To get a grip on process loops, you should rely on standardized process flows. In general, try to keep the process variance small, because the fewer different process flows there are, the fewer deviations can occur.

By standardizing and reducing process variance, you also create the prerequisites for the next step, process automation. Repetitive, standardized process steps can nowadays be automated by technical means, the keyword here is Robotic Process Automation (RPA). Celonis can also support you in this area with the included modules for process automation. Alternatively, you can also rely on well-known RPA providers such as Blue Prism or UiPath.

Advantage Process Mining

What is the final difference between process mining in order processing and classic process analysis? There are parallels and differences in the procedure: both approaches penetrate deep into the process flows and require intensive, explanatory support from the specialist department.

However, traditional process analysis cannot provide a comprehensive view of large product portfolios and potentials cannot be quantified without extensive data analysis. This is precisely where process mining develops advantages:

• Realistic: Mapping of actual processes from the main process to the exotic process
• Completeness: Consideration of the entire product portfolio instead of selected representatives
• Speed: Significant time savings thanks to the integrated approach of pure process flow analysis and data analysis
• Precision: Precise quantification of delivery time reductions and OTD gains down to the decimal place for the first time
• Sustainability: Process mining in live operation can be anchored in your own organization

I would particularly like to emphasize sustainability, as it differs from “normal” technological progress. Thanks to its connectors, once a process mining model has been set up, it is able to provide ongoing insights into the order handling processes without a great deal of effort. Process mining can be integrated into the day-to-day work of the operational organizational units and thus has the potential to become a valuable tool in their management and a noticeable optimization lever.

Summary

Does this mean that the classic process analysis of order processing has been sidelined? Not at all, but its role is changing. The strength of the proven methodology is the fact that the people involved develop a deep understanding of the process and product. And a process mining project cannot do without this knowledge. This closes the circle and it becomes clear: process mining is the future – but classic process analysis remains the basis. A successful project therefore combines both approaches in order to sustainably improve delivery times and adherence to delivery dates.