I think it is probably safe to say that most users don't use Tableau extracts as their primary means of data storage. Generally companies store their data in databases (either hosted on premise or in the cloud) that can be accessed by whatever front-end tool the end users want, whether that is Tableau, Excel, or something else.
Agreed. We also have all our data stored in a database (data warehouse). However, even with data stored in a proper designed database, there is always some logic (percentages of total, time series logic etc.) that is best calculated in cubes / front-end tools. Having different tools accessing the database directly, I need to define this logic in several places - which is something that I really don't want to do.
Of course. Ideally you want the metadata / business logic stored in a single location. Unfortunately compared to other BI solutions out there, Tableau doesn't offer much in terms of metadata management. Can you push the business logic back to the database level - maybe store it in some views / cubes that Tableau could access?
As I see it, views will not be an optimal place for percentages of total, time series logic etc.
Cubes are an option. They could ensure all business logic is stored in one place. The drawback being limitied flexibility when accessing the cubes from Tableau - providing arguments for increased use of Excel to solve certain tasks. Which is why I would prefer to avoid this option.
How would you go about this?