A clean Q3 beat, a 27% drawdown from the August peak, and the first quarter where LLM referrals actually nudged off zero
BKNG · Earnings Call · 2026-11-03
A beat the tape refuses to pay for
Booking Holdings did what it has done for eight straight quarters: it beat. Room nights grew 7% (up from 5% in Q2), gross bookings 10%, adjusted revenue 9%, adjusted EBITDA roughly $4.6 billion at a 47.0% margin, and adjusted EPS of $4.66 rose 17% — “room nights, gross bookings, adjusted revenue, and adjusted EBITDA all exceeded the high end of our guidance for the 3rd quarter” — Ewout Steenbergen, Executive Vice President and Chief Financial Officer · 2026-11-03. The operating leverage is real and visible in the filings: operating margin of 34.8% on $7.4 billion of quarterly revenue, up 8% year over year, and management is still handing back enormous amounts of cash, with $4.0 billion repurchased in the quarter alone, up 166% year over year.
And the stock, as of the report, sits down 14.7% over the last 90 days and roughly 27% below its August high. The gap between the operating result and the price is the entire story here, and it has a name: the horizontal personal agent. The September arrival of Meta's Muse — and the sharp re-rating in travel names that followed — convinced investors that the front door to travel booking might stop being a search box, and might stop being Booking's. Management came to this call knowing exactly which question it had to answer.
"Hasn't moved so much" becomes "accelerated"
Here is the change worth marking. For at least four consecutive quarters, the LLM-referral disclosure was the most boring line in the deck — a ritual disclaimer that this traffic was tiny and, crucially, static. In August it was flatly dismissive:
that is still significantly below 1% of our room nights. And that hasn't moved so much recently. So no material change over the last few months or quarters.
On Tuesday that language changed in kind:
LLM referrals have accelerated over the last several weeks, but still account for significantly less than 1% of Booking.com's room nights.
Both halves of that sentence matter. The absolute level is still noise; the direction has flipped, and Ewout Steenbergen tied it explicitly to model releases improving citation rates. A year ago Glenn Fogel was telling investors that the idea these agents would "cause this giant shift" was, in his words, “not the way the world is going to work” — Glenn Fogel, Chief Executive Officer · 2025-10-28. He hasn't abandoned that view — he's now trying to monetize it. The personal agent thesis Booking is selling is not defense, it's positioning: agents will need supply, real-time availability, payments and 40-plus languages of service, and the company that owns 4.7 million properties becomes the default rail. “They like to partner with aggregators. It's going to be very expensive for them to connect to 4.7 million properties directly and get all the information about real-time data, pricing, availability.” — Operator · 2026-11-03 Note the phrase "brand credibility" — the same pitch Booking has made to independents for two decades, now aimed at Meta.
This is not a lonely theme. The personal agents wave shows up across the global keyword set and across this week's reporters — Shopify with its Personal Agent Protocol, PayPal, DoorDash, Twilio, all circling the same horizontal agent construct, with even a "Grok Bot" reference surfacing in the market-wide top keywords. Booking is not riding an idiosyncratic idea; it is a large-cap at the center of a market-wide one. The tape, however, has so far treated that wave as a threat to OTAs rather than a tailwind for them.
Greenfield labs and a cost line that is actually moving
What is genuinely new — not a re-tread of the last four calls — is the willingness to build outside the mothership. Two AI-native brands launched inside six weeks: Lola in September, and Veo, a proactive travel agent that ingests a traveler's email history and keeps working after booking. Fogel frames them as deliberate testbeds: “it really is a wonderful place to be when you have the capabilities, the capital, the resources, the number of people that you can afford to do this type of agile creating new things” — Glenn Fogel, President and Chief Executive Officer · 2026-11-03. The early numbers are anecdotes, not evidence — Agoda's AI assistant users converting at roughly twice the rate of non-users and booking ~20% higher value — but they are at least measurable, which is more than the company offered in February.
The efficiency side is where the P&L evidence is strongest, and it recurs from prior calls rather than arriving fresh: customer service costs fell in absolute terms on higher volume, cost per merge request declined double digits, and Agoda's blended cost per million AI tokens fell more than 25% between June and September. Fogel made the point bluntly a quarter ago that AI is "everywhere except for in the P&L" — Booking's counter is that it has a line item where it shows up.
The guide is where the story actually lives
Strip out the AI theatre and you are left with a decelerating fourth quarter and a stock priced for one. Room nights are guided to +1% to +3%, adjusted revenue +0% to +2%, and adjusted EBITDA roughly flat — against a Q3 that grew room nights 7%. Management attributes it to the toughest comp of the year, October disruptions, and planned step-up in investment behind flights, attractions, dining and infrastructure. The only forward-looking reassurance offered was qualitative and hedged: on-the-books statistics for the first half of 2027 "look healthy," while 2027 guidance itself was deferred to February.
Two things deserve skepticism. First, the pattern of teasing without disclosure — Genius has been promised as something "not nearly what it can be" since at least February, re-teased in April, and re-teased again this week with the same "I don't want to give away too much." That is a multi-quarter pattern, not a new signal. Second, the one structural disclosure that is coming: a combined B2B business launching January 1, with segment numbers to be broken out. That will finally let investors see the incrementality management insists exists, and explains the marketing leverage optics — faster B2B rev-share growth flowing through the marketing line.
And notice what has quietly fallen out of the narrative: the Middle East conflict, a top-ranked keyword for the company in Q2 and a persistent 2-3 point guidance headwind through the spring, is no longer a headline theme even though guidance still assumes its effects persist through year-end. It has been absorbed into the baseline — alongside a balance sheet that now carries $2.8 billion of net debt, a deliberate consequence of the repurchase program.
The honest read: a high-quality compounder growing high-single-digits, de-rated by roughly a third on a technology question it cannot yet answer with numbers. Fogel's most perceptive remark had nothing to do with booking curves — it was the regulatory aside, that “there's parts of the world where the adoption of these horizontal personal agents may be much slower than is anticipated” — Glenn Fogel, President and Chief Executive Officer · 2026-11-03. If he is right, the multiple re-rates back. If he is wrong, the sub-1% referral line becomes the most important number in travel.