Problem decomposition is where you start deciding what you’ll need to do first, second, third, and so on.
The problem you want to tackle is probably not atomic in scope. By this, I mean that your problem or feature implementation probably does not involve changing a single line of code, or a single function. It probably involves understanding how one or more functions, classes, or pieces of code interact with a host of other data structures and functions. As a result, your solution may require changes in multiple places.
This suggests that you are going to need to develop a strategy… you will need to decompose your problem into its smallest parts. In doing so, you are forcing yourself to understand the system that you are working in, how the parts of the system interact, and how your changes will impact that system.