A dashboard is useful when someone decides differently because of it. Start with one question, such as: “Which open orders put this month's close at risk?”
Define the metric in writing: which orders count, which cut-off date applies, what happens with cancellations? Only then connect the data. Check missing values, duplicates and access rights. An attractive chart on incorrect data is worse than a plain, verified table.
Three elements are often enough for the first page: a trend over time, a breakdown by owner and a list of cases that need action. Show when the data was last refreshed. A scheduled refresh does not mean that the source always delivered on time or without errors.
Test the same question with management and staff. Can both explain the figure and get from the chart to the underlying case? If not, improve the data model and labels before adding new visuals.
Licensing depends on authors, readers, sharing method and Fabric capacity. Microsoft's current Power BI pricing is the starting point. Also budget for data maintenance and support.

