Hotel bookings in Power BI Service: data model, refresh and access control
Hotel dashboards are easy to draw; reliable booking measures require clean date logic, cancellation handling, refresh monitoring and per-property access boundaries.
Define the measure before drawing the dashboard
A booking count is not hotel occupancy. One reservation can span several nights, so booking rows must not be treated as occupied room-nights without a deliberate definition. Document the meaning of BookingId, PropertyId, StayDate, Status and RoomNights before building measures. Keep cancellations, no-shows and date changes distinct. A source-table row count is not automatically a management KPI.
Star schema and DAX filter context
Booking transactions can form the fact table while hotel and calendar attributes form dimensions. When BookingId repeats, decide whether the metric represents distinct bookings or room-nights. A filtered measure for confirmed bookings is a useful starting point, but evaluate it separately under property and period filters. Treat cross-filter relationships as assumptions to test rather than guaranteed behavior.
ConfirmedBookings = CALCULATE(DISTINCTCOUNT(Reservations[BookingId]), Reservations[Status] = "Confirmed")Refresh, data sources and row-level security
A visually plausible report does not prove a successful data refresh. If data comes from an on-premises network, verify gateway connectivity, credentials and the last completed refresh timestamp. For a property manager who must only see their own PropertyId, test row-level security with restricted identities. Report page filters are not a security boundary.
Historical context and verification
This is a present-day examination of reporting challenges relevant to 2021, not an article originally published then. Test a duplicate booking, a cancelled reservation and a stay that crosses a year boundary. Reconcile DAX measures against independent source queries. If totals differ, inspect relationships, date granularity, data types and refresh timing one by one.
Comments
No comments yet. Be the first to contribute.
✎ Write a comment