I do lots of investigative visualization working on development timescales like one hour, one day, one month or years. I especially like studying alien semistructured data formats where structure was intended but poorly executed and schema lost. A good example would be sense-making of some random xml or yaml file.
Say you encounter a large data file with sketchy details and a hint that it could be valuable. You have questions.
Now we consider how to reap the value available in this data over time and in the context of other things we know.
These and related skills have variations that address more specialized applications.