Assuming by clusters, you mean sections, then it really depends. I know this type of visualization as a kanban board, so you can also look that up.
By grouping all your tasks into sections, you can see work broken up and gain insight into the health of a task or project. In your example, the 3 sections are very basic view of the tasks that have been identified:
- Open - this would likely be the backlog of tasks, what do I know that I have to do but haven't started yet
- In Progress - what am I actively working on.
- Closed - what have I completed.
By viewing tasks grouped into these buckets, you can gain insight into your project health by visualizing how much work you haven't started, what your team is working on, and what is already done. If I've completed one task, am working on one task, and have a backlog of 100 tasks, I can tell very quickly that this project will take a long time. In addition, tasks in these sections should also include responsibility and duration at a minimum.
To determine the sections you are going to use, its really a question that the you can work through with your team. Again, the example provided in your link is basic, but it works if you need a basic visualization. You can get more complex by adding sections if it serves a purpose. So maybe adding another section that at the beginning called "Backlog", which are items you know about but haven't been looked at by the team. Then the "Open" section can be items that have been looked at by the team and scoped out, but not started. It really needs to be fit for purpose, which is a conversation best had with the team or set by a governance body like a PMO.