Process mining is a technique that combines business process management with data science. It comprises three main activities: process discovery, conformance checking, and process enhancement.

1. Process Discovery: This is the first step in process mining, which aims at discovering the actual process by observing events or activities in an information system. This part focuses on understanding the existing process.

2. Conformance Checking: It involves comparing a predefined process model to the discovered model in order to assess similarities and differences, thereby identifying deviations from the designed process model to the one that is actually executed.

3. Process Enhancement: After the discovery and the conformance checking stages, the next step is process enhancement, which involves improving and optimizing the discovered process. It aims to extend or improve the quality of the existing process model for better performance (e.g., correcting or automating particular tasks, removing bottlenecks). It might involve adding, removing, or changing steps in order to make the process more efficient. 

The key difference between process enhancement and the other two activities is that while process discovery and conformance checking are mainly about understanding the processes as they are, process enhancement is about using this understanding to improve and optimize the processes. Process enhancement takes a more forward-looking and proactive approach aimed at business improvement.