What is process mining?

Process mining discovers the real workflows hidden inside your conversations and shows them to you as a map. Instead of guessing how a ticket, order, or request actually moves from start to finish, Tekst reads your messages, identifies the steps involved, and draws the process the way it really happens - including the detours and variations you might not expect.

This article explains what process mining does and how it fits together. To build your first one, see the "Set up your first process" article.

What it does

When you create a process, Tekst analyzes a sample of your conversations and automatically detects the activities (steps) they contain - for example "Order received", "Question asked", or "Ticket closed". It then works out the order those steps happen in across many cases and draws a process map: a diagram of the steps and the paths between them, annotated with how often each path occurs and how long it takes.

The result is a clear, data-driven picture of how work actually flows, so you can spot the most common routes, the slow steps, and the variations worth investigating.

Where it comes from

Process mining builds on the rest of the Tekst platform:

• Your conversations provide the raw material. Each conversation (or entity) becomes a case that flows through the process.
• Your models help label the steps. Activities derived from message content come from classification, so the steps reflect what your models already understand about your messages. To learn more about models, see What are models on Tekst?.
• Your inboxes (integrations) define the scope. A process draws from the inboxes you point it at.

Key things you can do

Once a process is built, you can:

• Read the process map to see the steps and the paths between them.
• Explore variants - the distinct routes cases take through the process.
• Measure throughput time and set a target to track how long cases should take.
• Filter by inbox, topic, and time frame to focus on a slice of the data.

You can also clean up a discovered process by renaming, merging, hiding, and describing its activities so the map reflects how your team actually thinks about the work.

How it relates to analytics

Process mining and analytics are complementary. Analytics answers "how much and how often" across your messages; process mining answers "in what order and how long". For broader reporting, see the Analytics overview.

Where to start

To learn the vocabulary first, see the "Activities, variants, and the process map" article. When you are ready to build, see the "Set up your first process" article.