AI concierge and revenue management across 14 UAE hotel properties, lifting RevPAR 11%

Industry

Hotels

Market

UAE

Client

Hotel operator with 14 properties in the UAE

Sponsor

Director of Operations

What we built

AI concierge and revenue management

The situation

Running one hotel well is hard. Running fourteen to the same standard is a different problem. The client operates 14 properties across the UAE, each with its own layout, facilities, restaurants, policies and local quirks, and each staffed by teams that change over time. The Director of Operations described the core issue plainly: guest service consistency was impossible to maintain at that scale without something that knew every property as well as the best person at each one.

Alongside service, there was pricing. Rates were being set property by property, with varying levels of analysis behind them, which left revenue on the table in a market where demand swings sharply with events, seasons and competitor moves.

What we built

Two connected systems: a concierge that gives every guest the same quality of answer at every property, and a revenue management layer that brings the same discipline to pricing.

A concierge that knows each property. Guests ask the same things everywhere: check-in times, breakfast hours, pool access, spa bookings, restaurant reservations, transport, late checkout. The answers differ property by property. We built a knowledge base per property, maintained by that property's team, and a concierge that answers from the right one every time, in the guest's language, at any hour.

Requests that reach the right desk. Towels, maintenance, housekeeping and reservations are captured as structured requests and routed to the correct department at the correct property, so a guest's request is tracked rather than lost between shifts.

Revenue management with a human in the loop. The pricing layer brings together the signals a revenue manager weighs (pace, pickup, occupancy on the books, events, day of week, competitor positioning) and produces rate recommendations per property. Recommendations go to the revenue team for approval rather than straight to the channel manager, so the people accountable for revenue stay in control of it.

One view across fourteen properties. Operations and revenue leadership see service requests, response times and pricing decisions across the whole portfolio, rather than fourteen separate pictures.

How it went live

We did not launch fourteen properties at once. A pilot group of properties went first, with their teams building and correcting their own knowledge bases and the revenue team reviewing recommendations side by side with their existing decisions. Once the concierge's answers and the pricing recommendations had earned the teams' trust, the remaining properties followed on the same template.

Results

The group now runs a single, consistent concierge standard across all 14 properties, with requests tracked through to completion. On the revenue side, the Director of Operations reported that blended RevPAR improved by 11% in the first quarter, and that the return covered three years of the system's running costs.

What made the difference

Per-property truth. A concierge that gives a generic answer is worse than none. Each property owning its own knowledge base is what made the answers right.

Recommendations, not autopilot, on price. Keeping the revenue team as the decision-maker is why the recommendations were adopted rather than overridden.

Template, then scale. Solving the problem properly at a few properties first made the rest a rollout rather than fourteen projects.