On This Page
> resources > docs > intelligence > What data you need for process mining

What data you need for process mining

Process Intelligence Process Intelligence Available with a Process Intelligence seat or higher. Only seats on this plan or a higher one can use this feature. Compare plans Available with a Process Intelligence seat or higher

What Is Needed to Run Process Mining?

Process mining works with event log data: a record of what happened, when, and for whom. To analyze a process, you need three pieces of information for every step: a case identifier, an activity name, and a timestamp.

The Three Required Columns

Column What it is Example
Case ID A unique identifier for one process instance (one order, one ticket, one customer) 1001, PO-2024-0192
Activity The name of the step that was performed Create Purchase Order, Approve Invoice
Timestamp When the step happened, including date and time 2025-04-17 10:22:12

Optional columns add context: who performed the step (user/resource), how long it took, cost, or any other attribute you want to analyze.

Here’s an example of what a small event log looks like:

Case ID Timestamp Activity
1 10:00 AM Take Order
1 10:02 AM Prepare Lemonade
1 10:05 AM Serve Customer
2 10:03 AM Take Order
2 10:10 AM Resolve Customer Complaint
2 10:12 AM Prepare Lemonade
2 10:15 AM Serve Customer

Even this small amount of data is enough to reconstruct each case’s path and start answering questions about the process.

Why These Three Columns Matter

  • Case ID lets ProcessMind group events into cases, so it can reconstruct the full path of each process instance.
  • Activity defines the steps of your process. Clean, consistent activity names (always “Approve Invoice”, never “Approved” or “Invoice approved”) keep the discovered process readable.
  • Timestamp determines the order of events. ProcessMind needs it to sequence activities and compute durations, throughput, and waiting times.

Data Quality Basics

Process mining gives reliable results when the data is clean and consistent:

  • One row per event: each row in the log is a single event, not a summary.
  • Consistent case IDs: the same case always uses the same identifier.
  • No missing timestamps: an event without a timestamp cannot be placed in the process.

You do not need perfect data to start. ProcessMind works with the data you already have, and you can clean and prepare it as you go: see Data Cleaning & Preparation and Configuring Your Dataset.

Where to Get Data

Event logs are usually exported from the systems that run your process, such as ERP, CRM, ticketing, or finance systems. See Where to Get Data for system-specific guidance, or download ready-to-use sample logs from the Example Files.

Related Documentation