When working as a group or preparing to share a dataset, using a clear folder structure is absolutely necessary. But what makes a good folder structure?
First of all, this structure should be agreed to and adopted by all participants. This makes sure that it is coherent and understandable to all. Folder names should always be short and explicit. Users need to understand what files are within even without opening them. If your folder structure is complex due to the scope of the project, you should include some kind of documentation (probably a readme.txt) at the root of your folders to explain how they work.
The hierarchy of your folders should be consistent and logical. Go from a general, high-level folder (starting with a single folder for the project, using its name or acronym) to more specific lower-level folders. Your structure should not be too deep nor too shallow. Depending on the size of the project, this could mean 3-4 levels, but it could be more or less for small or very large projects.
For research projects, one option is to organise the folder levels based on research activity, data type, and kind of contents (publication, documentation, deliverables, etc.). But each project has its own needs and you will need to find out what works on a case-by-case basis. As part of your preservation strategy, it can also be useful to define temporary "temp" folders from which data can be safely deleted after usage.
This example from the UK Data Service presents one way your project folders could be structured:
You could do it the other way around: research activities could be the second-level folder, in which case they would contain their own data and documentation folders. The only requirement is that your structure is clear and coherent with the project.