We built an AI staffing tool for a hotel operator that combines live bookings with flight delays and local events to recommend shifts before demand arrives. Overtime fell 30%, and forecasts are 92% accurate.
Client: Pulse DynamicsIndustry: HospitalityService: AI & Automation
Overview
Client
Pulse Dynamics
Hotel operator with 26 properties worldwide
Engagement snapshot
The challenge
The client planned staff using fixed weekly rotas drawn up 14 days ahead. Those rotas could not react to live booking cancellations, last-minute VIP arrivals or local events moved because of the weather.
Staff burned out in busy spells and lost hours in quiet ones, so many left. Unplanned overtime cut into margins, and guests waited longer at the front desk and in the restaurants.
Staff burned out when shifts were short-handed and lost hours when they were overstaffed, which pushed turnover up.
Unplanned overtime was consuming 8% of total property EBITDA across the luxury portfolio.
Wrong staffing levels at front desks and food and beverage outlets at peak times lowered NPS scores and lost direct revenue.
The solution
We combined live booking data from the property management system (PMS) with outside signals, including airline arrival delays and local conference schedules.
AI assistants recommend shift changes to property managers 48 hours before occupancy peaks, so extra cover is in place before guests arrive.
Every automatic shift change is logged with a reason code, such as "Weather surge", which supports 100% labour law compliance.
Key features
An AI demand model uses bookings, flight delays and conferences to forecast occupancy and dining rushes.
Business impact
92% forecast accuracy
It recommends shift changes 48 hours before peaks and checks each one against working-time rules.
Business impact
30% less overtime
The AI watches local flight delays and city events to predict food and beverage rushes before guests reach the property.
Business impact
92% accuracy
Shift decisions protect fair treatment and rest hours for staff, which lowers turnover and meets union labour requirements.
Business impact
100% compliance
When demand spikes in one region, the system suggests moving staff in from nearby properties.
Business impact
30% overtime reduction
Before and after
Before
Fixed headcounts
Rotas based on historical averages that ignored live changes in bookings.
After
Staffing that follows demand
Staff numbers rise and fall through the day, based on checked demand data.
Before
Manager judgement
During staffing crises, managers went on instinct and labour laws were often broken.
After
Rules built in
Fixed rules block any shift that would break rest-period regulations.
Before
Fixing gaps by hand
Managers spent 15+ hours a week patching gaps in the schedule.
After
Automatic rosters
System-generated rosters cut manual planning time to under 30 minutes.
Results
30% OT Reduction
Planning shifts ahead removed the need for emergency cover at premium overtime rates.
Before: High-LeakageAfter: 30% Lower
92% Accuracy
Staffing now follows demand closely, so service holds up at peak times without overstaffing quiet periods.
Before: GuessworkAfter: Accuracy
15% NPS Lift
Staffing up early cut front-desk queues. NPS rose by 15%, which also lifted the properties' online review scores.
Before: +10%After: +15%
Testimonial
“Overtime is down 30% across our 26 properties. Shift recommendations arrive 48 hours before peaks, already checked against working-time rules, and the forecasts behind them are 92% accurate. Guest NPS is up 15% as well.”
Tyler DiazVP of Workforce Management, Pulse DynamicsKey results
Trust and control
Fixed rules check every AI-generated shift against the EU Working Time Directive and local union rules.
Compliance readyStaff personal details are removed and encrypted at the edge, and booking data is anonymised before any model uses it.
SecureThe cloud platform is built to keep performance steady across 500+ properties during simultaneous peaks, far more than the current 26.
Cloud nativeThe client owns 100% of the custom staffing models and reusable demand modules once the project is complete.
100% ownedFaster delivery
These modules saved about 8 weeks of setup, which we put into the PMS integration and the labour law rules.
Held each employee's role and home property, so the shift engine could match the right people to each gap.
Gave us a tested occupancy forecast, which we extended with flight delay and local event data.
Showed managers live staffing levels against demand at each of the 26 properties.
Tracked overtime spend against budget, so managers could see the effect of every shift change.
The platform combines live booking data from the property management system with outside signals, such as airline arrival delays and local conference schedules. From these it predicts busy periods, including food and beverage rushes, and recommends shift changes to managers 48 hours before occupancy peaks. Its forecasts are 92% accurate.
Overtime fell by 30% across the client's 26 properties. Before the project, unplanned overtime consumed 8% of total property EBITDA. Managers also spent 15+ hours a week fixing schedule gaps by hand, and automatic rosters now cut that planning work to under 30 minutes. Guest NPS rose by 15%.
Fixed rules check every AI-generated shift against the EU Working Time Directive, local union rules and rest-period regulations, and block any shift that breaks them. Each automatic change is logged with a reason code, such as a weather surge, so managers can show why a shift changed. Staff personal details are encrypted.
If overtime and short-staffed shifts are eating into your hotel margins, we can help you plan rotas from live bookings and local demand signals.