Hotels
Driving occupancy in a hotel trial.
46.6% off-peak occupancy uplift potential
The problem
Hotels are squeezed from both sides: rising operational costs that persist whether rooms are full or empty, and intensifying competition that pushes operators into discounting strategies that erode brand value and destabilise long-term pricing. The decisions that move the numbers — when to flex price, when to push promotion, when to staff up — depend on a clear read of demand. Most hotels do not have one.
Limited historical data and unreliable forecasting models make those calls harder, not easier. The incumbents in this space assume large, clean datasets and stable demand patterns — a poor fit for independents, new openings and secondary markets where demand is genuinely volatile.
What we built
We were tasked with predicting localised demand trends to increase occupancy during off-peak periods. The engine ingested the operator's booking data and modelled the impact factors — the environmental elements that move demand for bookings — alongside it, so each prediction came with the levers that drove it: occupancy, GOPPAR, upgrades, customer satisfaction.
The output is not just a forecast. It is a set of identifiable levers the operator can pull — which off-peak windows to target, which promotions to run, where staffing can flex — with confidence in why each one moves the number.
46.6%
off-peak occupancy uplift potential identified
3 months
pilot window
Bookings + impact factors
inputs
What it changed
The trial gave the operator four things it did not have before: a way to drive occupancy in off-peak windows without resorting to brand-damaging discounting, localised insight that was easy to understand and act on, growing internal confidence in how to use and communicate predictive technology, and a clear set of growth opportunities anchored to the behaviour drivers behind demand.