If you know your processes are underperforming but can't pinpoint exactly where things break down, you're not alone. Most organisations have the data; they just can't see it clearly. That's precisely what event log analysis changes.
An event log is the raw, unfiltered record of how work actually happens inside your systems. Not how it was designed to happen. Not how it looks in a flowchart. How it actually happens, step by step, case by case, timestamp by timestamp. For anyone serious about process mining, learning to read an event log isn't optional. It's the foundation on which everything else is built.
This guide walks you through event log structure, data quality, real-world examples, and the practical steps to go from raw data to genuine process intelligence.
An event log is a structured digital record of activities carried out within a system or process. Every row represents a single event. Every group of related events, tied together by a common identifier, represents one process instance, or case.
Unlike traditional reporting, which shows you aggregated summaries and designed-process assumptions, an event log captures operational reality. It comes from the systems your organisation already uses: ERP platforms such as SAP and Oracle, CRMs such as Salesforce, ITSM tools, ticketing systems, and custom databases. The data is already there. Process mining gives you the lens to read it.
Every valid event log for process mining must contain at least three fields:
1. Case ID
The unique identifier that groups all events belonging to a single process instance. Think: one invoice, one customer order, one insurance claim. The Case ID is the thread that holds the story together.
2. Activity
The named step or action performed at that point in the process, for example, "Invoice Received," "Payment Approved," or "Claim Escalated." Activities define the what.
3. Timestamp
When the activity occurred. Timestamps enable sequencing, duration calculations, and waiting time analysis. Without them, you have a list of events with no sense of order or pace.
Remove any one of these three, and process mining breaks down. The Case ID is the thread, the Activity is the story, and the Timestamp is the timeline. All three are required.
Once you have the three mandatory columns, richer analysis becomes possible through additional attributes:
The richer your attributes, the deeper your analysis. That said, the garbage-in, garbage-out principle applies firmly here. Only include fields that are consistently and reliably populated.
Reading an event log is simpler than it looks once you know the approach. Here's a practical method using a purchase-to-pay process as an example:
Once you can trace one case clearly, you're ready to look across all cases and identify variants, the different paths different process instances take through the same workflow. High variant counts often signal inconsistent execution or data quality issues, both of which are worth investigating.
Poor data quality produces poor process maps. Before you run any discovery algorithm, assess your event log against these criteria:
Once your event log is clean and loaded into a process mining tool, discovery algorithms, such as the Inductive Miner or Heuristic Miner, reconstruct your process as a visual map. Nodes represent activities; edges represent transitions between them, weighted by frequency and duration.
A high-quality event log produces a clear, readable process map. A messy one produces what practitioners call a "spaghetti model", an uninterpretable tangle of paths that obscures rather than reveals.
This is also where conformance checking becomes valuable: comparing your discovered process model against the designed reference model to surface deviations, compliance gaps, and rework loops.
Ready to analyse your first event log? Work through these steps:
An event log is not just a data file. It's a window into operational reality, one that most organisations have access to but haven't yet learned to read.
Master the structure, prioritise data quality, and the process intelligence follows.
At Verdant Data, we help organisations extract meaningful process intelligence from their operational data, from event log extraction through to actionable automation opportunities. If you're ready to move from process assumptions to process facts, get in touch with our team or explore our process mining resource library for your next step.
Get Updates and announcements from the Verdant Data Team
You can unsubscribe any time. Learn more about our Privacy Policy
.jpg)
%20(1).jpg)
.jpg)
.jpg)
.png)
