Directly-follows graph from the event log. Dashed red boxes are rework.
Activities
Cases that reach each activity, and how often it repeats.
What the map shows
- Every case runs intake → budget → quote → contract → requisition → PO; internal functions skip the contract.
- The longest waits sit in approval queues, not in the work: contract execution and the hand-off to purchasing.
- Three rework loops (intake returns, redlines, coding rejections) add a week or more each time they fire.
Variant explorer
Each variant is a distinct path through the process. Select one to see its cases.
All variants
Median wait by transition
Days between consecutive activities across the selected cases; the bar shows the median, the dark line the 80th percentile.
Throughput by dimension
Median days from intake submitted to PO sent, and the share of POs sent at least 14 days before the event.
| Group | Cases | Median days | P80 | Rework | On time |
|---|
Root causes found
Recommended workflow
The optimized intake-to-PO flow. Green steps are new or changed.
As-is vs to-be by step
Median wait in days for the selected cases, replayed with the improvements switched on.
Improvements
Switch each one on or off; every case is replayed with its own timings, so the gain is measured, not assumed.
Throughput distribution
Cases by days from intake to PO sent.
Contracting cases
Select a case for its timeline.
| Case | Category · stakeholder | Amount | Days | Rework | Lead time |
|---|
Synthetic event log from tools/process_mining_data.py: 126 contracting cases shaped by the category, stakeholder and amount mix of a real tracker, with no real records. The original analysis ran in a process mining platform; this page recreates its views.