Process Mining & Workflow Analyzer

The Prompt

This prompt performs comprehensive process mining analysis from event log data:

<process_analyst_persona>
You are a Process Mining Expert who has analyzed millions of process instances across ERP, CRM, and custom systems. You've identified $50M+ in efficiency gains by discovering hidden process variations and bottlenecks. You translate complex process data into clear visualizations that drive action.
</process_analyst_persona>

<mining_objective>
Analyze event log data to:
1. Discover the actual process flow (not the documented one)
2. Identify process variants and their frequency
3. Detect bottlenecks and delays
4. Find compliance violations and rework loops
5. Quantify automation and improvement opportunities
</mining_objective>

<event_log_data>
<log_format>
[PASTE YOUR EVENT LOG DATA]

Standard format:
| Case_ID | Activity | Timestamp | Resource | Additional_Attributes |
|---------|----------|-----------|----------|----------------------|
| PO-001  | Create PO | 2024-01-01 08:00 | jsmith | Department=IT, Value=5000 |
| PO-001  | Approve L1 | 2024-01-01 10:30 | mwilson | |
| PO-001  | Send to Vendor | 2024-01-02 09:00 | system | |
| PO-002  | Create PO | 2024-01-01 09:00 | alee | Department=Mfg, Value=25000 |

Required fields:
- Case_ID: Unique identifier for each process instance
- Activity: Step/event name
- Timestamp: When the event occurred

Optional enrichment:
- Resource: Who performed the activity
- Attributes: Value, department, priority, etc.
</log_format>

<process_context>
- Process name: [e.g., Purchase Order, Order to Cash, Hire to Retire]
- Expected steps: [list the "should be" process]
- Known issues: [current complaints or problems]
- Time period: [date range of data]
- Volume: [approximate number of cases]
</process_context>
</event_log_data>

<analysis_framework>
Execute comprehensive process mining with Python:

### Module 1: Data Preparation
- Parse event log
- Validate timestamps and sequences
- Handle missing values
- Create case-activity matrix

### Module 2: Process Discovery
Discover actual process flow:
- Extract unique activities
- Build transition matrix (activity → next activity)
- Calculate transition frequencies
- Identify start and end activities

Create process map showing:
- Activities as nodes
- Transitions as edges (thickness = frequency)
- Highlight most common path ("happy path")

### Module 3: Variant Analysis
Identify process variants (unique paths through process):

| Variant | Path | Frequency | % of Cases | Avg Duration |
|---------|------|-----------|------------|--------------|
| V1      | A→B→C→D | 450 | 45% | 2.3 days |
| V2      | A→B→E→D | 300 | 30% | 3.1 days |
| V3      | A→B→C→B→C→D | 150 | 15% | 5.8 days |

Flag:
- Rework loops (activity appears multiple times)
- Skipped steps (expected steps missing)
- Unusual paths (< 5% of cases)

### Module 4: Performance Analysis
Calculate for each activity:
- Frequency (how often it occurs)
- Service time (duration of activity)
- Waiting time (time between activities)
- Total cycle time contribution

Visualization: Activity timeline with waiting vs. processing

### Module 5: Bottleneck Detection
Identify bottlenecks by:
- Longest average waiting time
- Highest queue variability
- Resources with highest utilization
- Activities with most rework

Visualization: Heatmap of process showing wait times

### Module 6: Resource Analysis
Analyze by resource/person:
- Activities performed
- Volume per resource
- Average handling time
- Handoff patterns

Visualization: Resource-activity matrix

### Module 7: Compliance Checking
Check against expected process:
- Mandatory activities missing: [list]
- Sequence violations: [list]
- SLA breaches: [list]
- Authorization issues: [list]

### Module 8: Automation Potential
Score activities for automation:

| Activity | Volume | Standardization | Digital_Ready | Manual_Effort | Auto_Score |
|----------|--------|-----------------|---------------|---------------|------------|
</analysis_framework>

<output_deliverables>
Generate process improvement package:

### 1. Process Map (Visual)
- Discovered process flow diagram
- Edge thickness = frequency
- Node color = average duration (heat)
- Rework loops highlighted

### 2. Variant Overview
- Top 10 variants with frequency
- Happy path vs. deviation analysis
- Rework rate calculation

### 3. Performance Dashboard
- Overall cycle time distribution
- Bottleneck identification
- Wait time vs. work time ratio

### 4. Compliance Report
- Violations by type
- Cases with issues
- Trend over time

### 5. Improvement Recommendations
| Opportunity | Type | Impact | Effort | Priority |
|-------------|------|--------|--------|----------|
| Eliminate approval step X | Streamline | Save 1.2 days | Low | 1 |
| Automate data entry | Automate | Save 2 hrs/case | Med | 2 |
| Fix rework at step Y | Quality | -30% rework | High | 3 |

### 6. Automation Roadmap
- Quick wins (RPA candidates)
- Integration opportunities
- Redesign requirements

### 7. Complete Python Script
- Process discovery code
- Visualization generation
- Reusable for ongoing monitoring
</output_deliverables>

How to Use This Prompt

1. Extract event log: Pull from ERP, workflow system, or application logs
2. Ensure required columns: Case ID, Activity, Timestamp
3. Add context: Document expected process and known issues
4. Run analysis: Get discovered process map and insights
5. Prioritize actions: Use recommendations to plan improvements

Event Log Sources

Common sources for event log data:

• ERP transaction tables (SAP, Oracle)
• Workflow systems (ServiceNow, Salesforce)
• Database audit logs
• API call logs
• Manual tracking exports

Process Mining Follow-Ups

• “Deep dive into the 15% of cases with rework—what’s causing it?”
• “Show me the process behavior by department”
• “Compare Q1 vs Q4 to see if the new process is working”
• “Which activities could be automated with RPA?”
• “Calculate the cost of the current process vs. ideal state”