Create a Process Improvement Plan from the Ground Up With Process Mining

Before jumping in head first with process mining to identify and fix a broken or inefficient process, start with a strong plan to avoid headaches for yourself and others. A written plan provides direction, while giving your stakeholders something to react to. They’ll see exactly what you want to do, and have the opportunity to make suggestions.

Include these five phases of process improvement in your plan:

• Define
• Prepare
• Analyze
• Improve
• Monitor

We’ll define these phases below, and break them out into different elements. The poster below walks through these steps, or you can download a PDF version of the process improvement plan here.

As is true with any great plan, you have to start by identifying where you are and where you want to go. In the Define phase, begin by defining the problem, goal, stakeholders, and critical process output. As you work through this information-gathering stage, you’ll want to identify the invested parties and learn about their difficulties with the process. Find out what they’re trying to achieve and why, and where their roadblocks are.

1. Process selection.
In this step, decide what process you want to improve and why. Consider the following criteria:

• Data availability: Do you have enough process data with time stamps available, so that you can conduct meaningful analysis?
• Management: Is there support from management to pursue this process? Are sufficient resources available?
• Potential: Where can you make the biggest improvements? What business value will you deliver by focusing on this process? Which process offers the most added value for customers?

2. Relevant stakeholders.
Ensure that all process participants are involved in the project. When it’s time to redesign your process, who will be involved? Develop a RACI (responsible, accountable, consulted, and informed) model for staffing your process improvement plan and identify these stakeholders:

• Process owner and process manager.
• Data owner.
• Specialists and system experts.
• Decision makers.

3. Process KPIs.
Define the KPIs relevant for the analysis in your business domain to ensure you’re capturing the right data for analysis. While some KPIs are generic, like return on investment (ROI), you’ll likely want to get more specific with metrics that are relevant to the department. Here are a few common process KPIs to evaluate:

• Processing times.
• Number of process runs.
• Number of process variants.
• Error rate.

4. Process-related questions.
Specify the question or questions to be answered in the process analysis.

• How good is the process performance? Process mining analysis can shed light on processing times, number of process runs and variants, and error rate to help you evaluate process performance.
• How standardized is the process? Is it performed the same way every time, or are there multiple ways to execute the same process?
• How automated is the process? Automating manual, highly repetitive activities where it makes sense can reduce costs, improve efficiency, and free up time for higher value work.
• Does the process need to comply with internal or external regulations? You can use process mining to compare your discovered model to your target model, with all the compliance needs mapped out, to identify where the process may be falling short.
• What is the optimization potential? Even small process improvements can have a far-reaching impact on the business: better user experiences, faster time to market, increased customer satisfaction, and improved compliance.

5. Creation of a target model.
Determine whether a target model of the process already exists.

• If no model exists yet, you can create it in this step manually or generate it in Step 8.
• A target model represents the ideal state of your process, or what you hope to achieve from your improvement efforts.

The target model can be used as the basis of comparison once your actual process has been mapped.

In the Prepare phase, you'll need to extract and transform the data for analysis.

6. Data and data sources.
Check which process data is available and ready for use. Where is it located? How can it be used? Consider the following:

• Data type/format.
• Storage location: system/database.
• Mapping of time stamps to activities.

7. Data extraction and transformation.
Next, you’ll need to ensure data is in the correct format. For analysis to work, data must be prepared as event logs. Check whether the log files contain at least the following: case ID, activity name, start and end times. Including additional attributes is optional. Mining Prep, a module of Appian Process Mining, simplifies data preparation with a no-code tool for extracting and transforming.

Using data science and machine learning, process mining analyzes system log files to create clear and concise maps of your processes. This replaces subjective and costly interviews with facts—a far better way to document how work happens. The data-driven approach provides insight into what people, systems, and organizations are actually doing, as opposed to what you think they’re doing.

8. Data import and processing.
Now you’re ready for analysis.

• Load the transformed data into the process mining tool. It automatically generates a model of the actual process (this is called the discovered model).
• If no target model was available in Step 5, create a model of the target process in the process mining tool.

In the Analyze phase, you can start to pinpoint problem areas in your process. Use the discovered model that process mining created for you in Step 8 to highlight optimization potential. This visual helps you quickly identify inefficiencies, like bottlenecks, rework, and other activities that stall progress.

9. Model enhancement.
Model enhancement is where you dig into how your actual process works by analyzing your discovered model. Analysis reveals details about the following factors:

• Process duration.
• Frequency of unexpected process steps.
• Distinctive process sequences.

10. Conformance checking.
In conformance checking, the discovered model is compared with the target model (Step 5 or Step 8) to detect deviations and check process conformity. For example, you may find that users are skipping or duplicating process activities or executing unplanned activities that pose compliance risk.

11. Root cause analysis.
Root cause analysis helps you better understand process deviations. Use it to identify problematic attributes, patterns in deviations, and indicators for effective optimization. In Appian Process Mining, you can use root cause analysis to understand the why behind process bottlenecks and inefficiencies, so you can build new workflows with confidence.

12. KPI analysis.
Evaluate your process data using easy-to-understand dashboards that provide visualizations of the relevant KPIs (Step 3). You’ll see where the process is falling behind, and learn where to focus your efforts.

In the Improve phase, you’re finally ready to develop and implement solutions. Building your new process on a unified, enterprise-grade, low-code automation platform will help you move quickly and efficiently. Optimize workflows with AI-powered intelligent document processing (IDP) to quickly process documents for teams; use business rules to continually replicate the right procedures; and implement robotic process automation (RPA) to save employees time on repetitive, high-volume tasks.

13. Addressing the issues.
Based on your findings, determine which process improvement/optimization measure best suits your needs. Examples include:

• Business process automation.
• Additional staff training.
• Greater standardization and harmonization.
• Implementation of new/additional IT systems.
• Adaptation and optimization of process flaws.

14. Solution implementation.
Work with the team and all relevant stakeholders (Step 2) to put your solutions into practice. Remember to give each invested party a chance to review, suggest changes, and buy in before you begin. Provide the “what’s in it for me” angle for each stakeholder. What about this new process will help them reach their goals faster or more effectively than before?

15. New target model.
Replace your earlier target model (Step 5 or Step 8) with the current, optimized process. This will be your new target model and serve as the basis for comparison during future optimization efforts.