Best advice i can give you is that you just store the data raw and then create measures to calculate whichever way you need.
For example, you might simply store your invoices, with their amounts and dates. You can then create sets of measures with different time intelligence to consider different scopes.
Power BI is context driven. So if you create a measure to perform a sum, you can reuse that measure in many different contexts to constrain what the limitations of the is. This is what is meant be "context"
To put it another way: You can choose to sum invoices, or maybe you want to know what the largest value invoice is. This is how you wish to aggregate your data, which some charts and graphs let you do seemlessly. You can also write measures so you can call this aggregation in later measures and functions.
Next you might say "Yeah sure thats a sum of invoices, but over what period of time do you mean?" The time period is the context. So if you take a graph, make the date the axis, then the plot points each have a specific context which Power BI runs your measure on. So if your graph axis was monthly, you would get a data point that sums up invoices for that specific month.
So you set your lowest level measures on what calculation you are trying to perform, and your graphs, charts, matrixes and later measures set the context. As you learn more and more you'll see how you can build outwards to some super cool and tricky stuff.
