Labor that breathes with the season
A Bali resort running three F&B outlets on Mews cut low-season F&B labor-to-revenue by 14%, outlet-level demand forecasts consolidated outlets on forecast-quiet monsoon days, while high-season service stayed as it was. Guest satisfaction held through both seasons.
Measured results, property anonymized at client's request. Metrics verified against an agreed baseline.
The problem: APAC's lowest F&B revenue per room, staffed like peak week
APAC hotels earn the region's lowest F&B revenue per available room (~US$43.90), which makes idle outlet hours expensive. In monsoon January the resort ran ~48% occupancy, yet the beach grill, lobby café and main restaurant all kept full brigades, because the roster was a copy of December's.
What changed
Outlet-level demand forecasting is the prediction of covers per restaurant, bar or café per day from the bookings and in-house guests a property system already holds, so each outlet is staffed to its own expected demand.
Mews reservation and covers data became per-outlet demand forecasts. On days two outlets forecast under 30% capacity, the schedule consolidated service into one, staff rotated between outlets, and guests saw the change only as a shorter walk. How hotel demand forecasting works →
The data behind this chart
| Month | Before (%) | After (%) |
|---|---|---|
| Jan | 41 | 35 |
| Feb | 42 | 36 |
| Mar | 36 | 33 |
| Apr | 33 | 31 |
| May | 31 | 30 |
| Jun | 29 | 29 |
| Jul | 27 | 27 |
| Aug | 26 | 26 |
| Sep | 29 | 28 |
| Oct | 32 | 30 |
| Nov | 36 | 32 |
| Dec | 30 | 29 |
“Low season used to mean watching payroll eat the P&L. Now the quiet weeks cost what quiet weeks should cost.”
HotelCadence is workforce software for hotels. It reads the bookings already in your property system, forecasts how busy each department will be hour by hour, builds the roster to match, records worked hours and prepares payroll. Department heads approve every roster. From US$6 per scheduled employee per month.
Where does a 14% low-season saving come from?
January-February occupancy dropped to ~48% but all three outlets kept full rosters. Outlet-level covers forecasts showed two outlets running below 30% of capacity on weekdays, consolidating them on forecast-quiet days cut paid hours while total resort covers were still served.
Why did high season stay untouched?
The optimizer schedules to the service standard first. In July-August the forecast demanded full rosters across all outlets, so it rostered them, and the labor-to-revenue line stays almost flat in peak months because it was already close to right.
How do OTA-heavy bookings affect this?
Indonesian resort bookings are OTA-dominated (77.9% of reservations) and short-lead, which makes a month-out gut forecast wrong more often. Booking-driven forecasts from Mews re-read the book daily, so quiet days are called days ahead of service.