Booking Holdings Reports 10% Cut in Support Costs Per Reservation
Booking Holdings cut customer service costs by roughly 10% per reservation while increasing bookings by about 10%, according to CFO Ewout Steenbergen during the company's fourth-quarter earnings call. The company achieved the dual improvement by deploying generative AI assistants across its travel platforms, including Booking.com, Priceline, OpenTable, and Agoda.
The results matter because most companies struggle to extract both cost and revenue gains from AI spending. A PwC survey found only 12% of CEOs report achieving both cost reductions and revenue benefits simultaneously.
What Booking Built
Priceline's AI travel agent "Penny" is showing higher engagement in early tests. The agent helps users plan trips and make bookings with less human intervention.
Booking.com added enhanced natural language search and automated post-booking service flows. These features reduced the number of support contacts customers needed to make.
OpenTable deployed an AI Concierge that answered 80% of diner questions when it launched in July 2025. The system handles routine inquiries about reservations, menus, and restaurant policies.
Agoda reported double-digit year-over-year reductions in customer service cost per booking, though the company did not disclose the exact percentage.
How This Works Technically
Achieving a 10% cost reduction requires careful system design. Travel companies typically combine conversational agents with retrieval-augmented generation-a technique that pulls accurate information from booking databases-and automation for routine workflows.
The key is routing: determining when to let an AI agent handle a request and when to escalate to a human. Poor routing creates friction. Errors on booking-critical tasks damage both customer satisfaction and margins.
Companies measure success through cost per interaction, escalation rates, task completion rates, and whether automated handling increases or decreases bookings. Booking's 10% cost reduction alongside higher booking volume suggests the company got the routing right.
What Happens Next
Watch whether Booking sustains these cost reductions as customer volume grows. Efficiency gains sometimes erode at scale when edge cases multiply and error rates climb.
Customer satisfaction and escalation metrics will matter more than raw cost numbers. If AI agents are cutting costs by refusing to help or by making mistakes that frustrate users, the savings disappear when those customers stop booking.
For support teams, the lesson is straightforward: AI works best on high-volume, routine tasks where the cost of a wrong answer is low. Booking's success with reservation confirmations and policy questions reflects that reality. More complex issues still need humans.
Learn more about AI for Customer Support and AI Agents & Automation to understand how these systems work in practice.
Your membership also unlocks:
