"The analysts responsible for each business unit must see all the data the board sees, except the profit and loss data, which must be restricted to only their business unit's data. The analysts must be able to build new reports from the dataset that contains the profit and loss data" => one dataset and two separate workspaces Reason: All data can be imported into one dataset also if these are two logical models. Shared dimensions can be reconsumed in both models. Reports and additional materials can be shared to the board with an app. The "profit and loss" data model needs RLS for the analysts and the analysts must have just read access to the original workspace. In a separate workspace with contributer (or more rights) they can create new reports (with live connection to the dataset). It is also stated that the new reports mustn't be shared so therefore no need to include them into the app. Import vs. DirectQuery: Due to RLS requirements an imported dataset is needed. It is not possible with file sources and Sharepoint lists.