Booking Holdings CFO Sees Minimal AI Referral Risk, Developing Agentic Tools to Boost Direct Traffic
Booking Holdings CFO said large language models generate negligible and stagnant referral traffic, with direct traffic averaging around 65% in 2025. The company is developing internal agentic AI tools across platforms like Priceline’s Penny, Booking.com, and Agoda to counter potential traffic losses and match horizontal AI offerings.
1. Negligible AI Referral Traffic
CFO Ewout Steenbergen said referral traffic from large language models remains very small and sees no clear growth trend, indicating minimal immediate disruption from AI-driven search to Booking's platforms.
2. Direct Traffic Risk and Strategy
While disintermediation risk is low, Steenbergen highlighted a more realistic concern is preserving direct traffic, which comprised mid-60% of consumer bookings in 2025, prompting initiatives to maintain and grow this channel.
3. AI Tool Development Across Brands
Booking has two internal startups building agentic tools for trip planning and fulfillment, with Priceline’s Penny already operational and Booking.com and Agoda developing similar AI features to compete with horizontal platforms.
4. LLM Limitations Highlight Complexity
Steenbergen emphasized that major LLMs cannot handle fulfillment tasks like cancellations, customer service, payments, tax compliance or regulatory requirements, underscoring the operational complexity that supports OTA relevance.