Booking Holdings
Q2 2026 (reported August 4, 2026) — room nights 325M (+5%, above guidance), gross bookings $51.0B (+9%), revenue $7.4B (+8%) and adjusted EPS $2.54 (+15%), with $3.7B of buybacks; a 25-for-1 stock split took effect April 2. Q3 guidance of 3-5% room-night growth reflects Middle East conflict pressure on long-haul travel, and full-year guidance stays at high-single-digit growth. The agentic-disruption thesis is not in the numbers yet. The CFO puts traffic from LLMs, paid and unpaid, at 'significantly below 1%' of room nights with little change over recent quarters, and the company is one of the first partners on Google's agentic hotel reservations and part of OpenAI's CPC ad test. The risk is shifting from bypass to toll, as AI channels push traffic toward paid placement (marketing expense +11%, faster than gross bookings) while SEO erosion squeezes free direct traffic. Generative AI cut customer service cost per transaction by double digits, and the transformation program savings target rose to about $650M.
Scenarios
Scale makes the company a partner of AI agents rather than a casualty — an early partner on Google's agentic hotel reservations, in OpenAI's CPC test group, with LLM-sourced room nights still well below 1%. Merchant gross bookings at about 73% of the total (+4 pts YoY) and Connected Trip transactions growing at more than twice core transaction growth deepen payments and multi-vertical lock-in that a thin agent layer cannot easily replicate. AI-driven customer service savings and the ~$650M transformation target support margins.
If AI assistants become the default travel front door, OTAs buy traffic they used to get free — marketing already grows faster than gross bookings (+11% in Q2 2026) and SEO erosion is a stated headwind. Agents that reach hotel inventory directly would compress a take rate near 14.5% of gross bookings (Q2 revenue $7.4B on $51.0B). Near-term demand is soft (Q3 room-night guide 3-5%), alternative accommodations lag at +4%, an EU court upheld the block of the Etraveli flight acquisition, and the CEO concedes Connected Trip does not yet work as advertised.
Key Factors to Watch
- ●Q2 2026 (Aug 4) — room nights 325M (+5%), gross bookings $51.0B (+9%), adjusted EPS $2.54 (+15%); FY2026 guide high-single-digit growth
- ●LLM-sourced traffic 'significantly below 1%' of room nights and flat over recent quarters (CFO Aug 4; CEO repeated it Sep 22)
- ●Early partner on Google's agentic hotel reservations and in OpenAI's CPC ad test — intermediated by AI, not yet bypassed
- ●Marketing expense +11% outpacing gross bookings — AI and SEO shifts push traffic toward paid channels
- ●Generative AI cut customer service cost per transaction by double digits; transformation savings target raised to ~$650M run-rate by end-2027
Score History
| Date | Score | Direction | Note |
|---|---|---|---|
| 2026-10-10 | 5.9 | Neutral | Text refresh to Q2 2026 (Aug 4), dims unchanged. Removed unverifiable claims (Microsoft/OpenAI 'Travel GPT' bypass, $700M agentic reinvestment plan, 560M monthly visits, '2026 guidance raised to 9%') and noted the 25-for-1 split (Apr 2). Monitoring dr 8 to 7. LLM share below 1% and partner status with Google/OpenAI argue lower risk, but the evidence is company-sourced only and paid-channel cost growth argues the other way. Gate fails on source independence and temporal stability. |
| 2026-04-10 | 5.9 | Neutral | Score 6.1→5.9 (dr 7→8). AI agent bypass threat materialized from theoretical to active market pricing. Microsoft/OpenAI Travel GPT in production; market repriced stock -8% on earnings beat. Direction positive → neutral. |
| 2026-03-08 | 6.1 | Positive | Score 6.9→6.1 (reweight + are 7→6). are corrected: AI enhances search/pricing but core revenue is travel commissions |
| 2026-03-08 | 6.9 | Positive | Score 7.1→6.9 (dr 6→7). External research cross-ref: highest disruption risk in batch — Google/OpenAI existential threat to OTA model |
| 2026-03-08 | 7.1 | Positive | Initial assessment from batch 4 research |
Customer Service Peers
Last researched: 2026-10-10
This is research and analysis, not financial advice. Scores reflect AI impact potential, not investment recommendations.