How to Graph Process Mining Results in Microsoft Power BI

Process mining transforms raw event log data into visual representations of how business processes actually execute, revealing bottlenecks, deviations, and optimization opportunities. Power BI provides the ideal platform for graphing and analyzing process mining results, combining its powerful data modeling engine with rich visualization capabilities. At EPC Group, we have implemented process mining dashboards for manufacturing, healthcare, and financial services clients, helping them achieve measurable process improvements.

What Is Process Mining and Why Visualize It in Power BI?

Process mining is a discipline that uses event log data from IT systems to reconstruct and analyze actual business processes. Unlike traditional process modeling, which documents how processes are designed to work, process mining reveals how they actually work in practice. The gap between the two is often significant and costly.

Event logs typically contain three essential fields: a Case ID (identifying a specific process instance, such as an order number or patient encounter), an Activity Name (the step performed, such as "Order Received" or "Invoice Sent"), and a Timestamp (when the activity occurred). Additional attributes like resource (who performed the activity), cost, and department enrich the analysis.

Power BI is uniquely suited for process mining visualization because it can handle the large volumes of event log data typical in enterprise environments, its DAX language enables complex time-based calculations required for process metrics, and its interactive visualizations allow stakeholders to explore process flows dynamically. When combined with tools like Microsoft's Process Advisor (now part of Power Automate) or third-party process mining tools, Power BI becomes the presentation and analysis layer for process intelligence.

Preparing Event Log Data for Power BI

Before you can graph process mining results, your event log data must be properly structured. Use Power Query in Power BI Desktop to import and transform raw event logs from source systems. Common data preparation steps include:

• Standardizing activity names - Event logs from different systems often use different names for the same activity. Create a mapping table that normalizes activity names (e.g., mapping "PO_Created", "Purchase Order Created", and "Create PO" to a single standard name).
• Handling duplicate timestamps - Some systems log multiple events at the same timestamp. Add a sequence number or use a secondary sort field to establish the correct order of activities within each case.
• Calculating duration metrics - Add calculated columns for duration between consecutive activities within each case. Use Power Query's Group By and custom column features to compute elapsed time, waiting time, and processing time for each activity pair.
• Filtering noise - Event logs often contain system-generated events, error retries, or administrative activities that obscure the core process flow. Filter these out or categorize them to keep the process graph readable.
• Creating variant classifications - Group cases by their complete activity sequence (variant) so you can identify the most common process paths and analyze deviations from the standard flow.

The data model should include at minimum an Events fact table (Case ID, Activity, Timestamp, Resource), an Activities dimension table (Activity ID, Activity Name, Category), and a Cases summary table (Case ID, Start Time, End Time, Total Duration, Variant, Outcome). This star schema enables efficient DAX calculations for process metrics.

Building Process Flow Visualizations

The core visualization in process mining is the process flow graph, which shows activities as nodes and transitions between activities as directed edges. Power BI does not include a native process flow visual, but several approaches work effectively:

Sankey Diagrams. The Sankey diagram custom visual from the AppSource marketplace is excellent for showing the flow of cases through process steps. The width of each flow indicates volume, making it immediately clear which paths are most common and where cases diverge from the expected flow. Configure the source as the current activity, the destination as the next activity, and the weight as the count of cases following that transition.

Decomposition Trees. Power BI's built-in Decomposition Tree visual allows users to explore process variations by drilling down through activity sequences, identifying which combinations of steps lead to specific outcomes (delays, errors, successful completions). This is particularly powerful for root cause analysis.

Custom Visuals for Process Maps. Dedicated process mining visuals such as the Process Map visual or third-party custom visuals can render process graphs with frequency and performance annotations directly on the edges and nodes. These visuals read the event log data and automatically generate process maps within Power BI.

Matrix and Heatmap Visuals. Use a matrix visual to create a transition matrix showing the frequency of transitions between every pair of activities. Apply conditional formatting to create a heatmap that highlights the most common transitions in darker colors and rare transitions in lighter colors.

Key DAX Measures for Process Mining Analysis

Effective process mining in Power BI requires purpose-built DAX measures that calculate process-specific metrics. Here are the essential measures our consultants implement for every process mining dashboard:

• Average Case Duration - Calculate the average time from the first event to the last event for each case. This is your primary throughput metric and the baseline for measuring improvement.
• Activity Frequency - Count how often each activity appears across all cases. Activities with unexpectedly high frequency may indicate rework loops or error correction steps.
• Bottleneck Identification - Calculate the average waiting time before each activity. The activity with the longest preceding wait time is your primary bottleneck. Use AVERAGEX over cases to compute this accurately.
• Conformance Rate - Compare actual process variants against the designed (ideal) process flow. Calculate the percentage of cases that follow the expected sequence versus those that deviate. Low conformance rates indicate process compliance issues.
• Rework Rate - Identify cases where the same activity is performed more than once. High rework rates for specific activities indicate quality issues at that process step.

These measures should be placed on dedicated dashboard pages with slicers for time period, department, resource, and process variant. This allows stakeholders to drill into specific segments and compare performance across teams, time periods, or customer segments.

Integrating with Microsoft Process Advisor

Microsoft's Process Advisor (now integrated into Power Automate) provides a purpose-built process mining tool that connects natively with Power BI. Process Advisor can ingest event logs, automatically generate process maps, identify bottlenecks, and recommend automation opportunities using AI. The results can be exported to Power BI for deeper analysis and broader distribution.

The integration workflow is straightforward: upload event logs to Process Advisor, let it analyze the process and generate the initial process map, then export the analyzed data to Power BI where you build custom dashboards tailored to your stakeholders. This approach combines Process Advisor's specialized mining algorithms with Power BI's superior visualization and sharing capabilities.

For organizations using task mining in addition to process mining, Process Advisor can record user desktop activities and incorporate them into the process analysis. This captures the detailed steps within each high-level activity, providing end-to-end process visibility that is invaluable for identifying automation opportunities.

Why Choose EPC Group for Process Mining in Power BI

With 29 years of enterprise consulting experience, EPC Group brings deep expertise in both process optimization and Power BI analytics. Our consultants understand the operational context behind process data, which means we do not just build pretty visualizations; we deliver actionable insights that drive measurable process improvements.

We have implemented process mining solutions across manufacturing (order-to-cash, procure-to-pay), healthcare (patient flow, claims processing), and financial services (loan origination, trade settlement). Our approach combines technical Power BI expertise with industry-specific process knowledge, ensuring that dashboards surface the right metrics for your stakeholders and compliance requirements.

Frequently Asked Questions