The Guide Everyone Hated Is a $700 Million Investment Decision: Upgrading Booking Holdings to Outperform

Results vs. Consensus

Q4 2025 Scorecard

Two providers scored the earnings line differently, one calling it a 1.2% beat and one calling it in line, so we present the range rather than pick the flattering read. Revenue is unambiguous. The more meaningful bar is the company's own guidance, which the quarter cleared on all four metrics for the second consecutive quarter.

Versus the Company's Own Guidance

Year-Over-Year Comparisons

Full Year 2025

The GAAP full-year decline is the one figure that needs explaining and is entirely non-operating. Net income fell 8.1% and GAAP EPS fell 4.1% because 2025 absorbed $1,380 million of unrealised foreign-currency losses on Euro-denominated debt, the $457 million KAYAK impairment, $205 million of transformation costs and $360 million of convertible-note charges. Operating income, the cleaner GAAP measure, rose 16.8% to $8,825 million.

Revenue

Revenue of $6,349 million grew 16.1%, beating the high end of guidance by about four points. Constant-currency growth of about 11% is the number to hold, and it compares with 8% in Q3 and 12% in Q2. The composition improved as well: management attributed the revenue outperformance running slightly ahead of the gross-bookings outperformance to higher-than-expected payments revenue, which is a structurally attractive source.

Merchant revenues grew 27.4% to $4,249 million and agency revenues fell 3.9% to $1,791 million, extending the migration to a payments-led model. For the full year, merchant gross bookings reached $130 billion, up 25%, representing roughly 70% of total gross bookings against about 63% in 2024.

Assessment: An 11% constant-currency quarter, three points above the long-term ambition, with revenue outgrowing gross bookings on payments mix. This is materially better than the Q3 print and it did not depend on currency to look good.

Margins

Adjusted EBITDA margin of 34.6% expanded 79 basis points, and the full-year figure of 36.9% expanded 193 basis points. That full-year expansion is the more important number because it was achieved while reinvesting roughly $170 million above baseline. The engine is fixed-cost discipline: full-year adjusted fixed operating expenses grew 7% as reported and about 4% in constant currency, six points slower than revenue.

Marketing went the other way in Q4, rising 22.3% and taking marketing to 4.5% of gross bookings from 4.2%. Three causes were given: opportunistic performance-marketing spend at attractive returns, a 13% increase in social-media investment, and brand marketing that was back-weighted in 2025 versus a 2024 pattern concentrated around the summer football championship. For the full year, marketing was 4.4% of gross bookings, essentially flat with 2024.

Assessment: The margin story is intact and the Q4 marketing de-leverage is defensible, because the spend bought a three-point room-night beat in the same quarter. The full-year figure being flat is the relevant fact; a single quarter of 25 basis points is noise inside a year that expanded margins by 193.

EPS and Cash Flow

Adjusted EPS of $48.80 grew 17.4%, trailing the 18.8% Adjusted EBITDA growth because of a higher in-quarter tax rate that management attributed to the timing of certain adjustments, partially offset by the 3.4% lower share count. The full-year tax rate was slightly below 2024, so this is a phasing item rather than a structural change.

Cash flow resolved the concern we raised last quarter. Q4 free cash flow of $1,416 million more than doubled, and full-year free cash flow of $9,086 million grew 15.1% against 19.6% Adjusted EBITDA growth. The gap between cash and earnings growth that looked worrying at the nine-month mark (6% against 20%) closed to a normal four-point spread by year end, confirming it was the deferred-bookings calendar rather than earnings quality.

Assessment: Free cash flow of $9.1 billion against $9.9 billion of Adjusted EBITDA is a 91% conversion rate, and the company returned $8.2 billion of it. That is the profile of a mature compounder doing exactly what it should.

Segment Performance

Revenue by Type

Merchant and the Payments Platform

Merchant gross bookings of $130 billion for the year grew 25% and now represent roughly 70% of the total, up seven points in twelve months. The CEO's defence of the business against language-model competition rested substantially on the operational depth this creates, and the specifics he gave are the most concrete articulation of the moat the company has offered.

"We have over 100 different payment methods, more than 50 currencies. Now why is that? Because the supplier has no idea how to take these strange payments that the customer wants to give." — Glenn Fogel, CEO

Assessment: Payments revenue was named as the reason revenue outperformance exceeded gross-bookings outperformance this quarter and as the reason full-year take rate rose. It is now both a margin contributor and the company's primary structural argument against disintermediation, which makes the 70% merchant mix one of the two or three most important numbers in the model.

Agency

Agency revenues fell 6.5% for the year and agency gross bookings fell 8.7% to $56.1 billion. The decline continues to look like migration rather than share loss, since merchant grew 25% over the same period and total gross bookings grew 12.4%.

Assessment: Unchanged from our prior read. The migration explanation holds as long as merchant growth stays well ahead of the agency decline, which it did by a wide margin in every quarter of 2025.

Advertising and Other

The KAYAK and OpenTable line grew 14.0% in the quarter and 11.3% for the year. No further impairment was recorded after the third-quarter $457 million charge, and management offered no update on KAYAK's trajectory.

Assessment: The absence of a second writedown is mildly reassuring and the absence of any commentary is not. One quarter of silence after a large impairment is normal; two would start to look like avoidance.

Key KPIs

Assessment: Almost everything improved, with two exceptions worth naming. Alternative accommodation growth slowed to about 9% in Q4 from about 10% for the year, and B2C direct mix was described as "similar to last year" on a full-year basis, having been described as up year over year on a trailing basis at Q3. The Genius mix moving from mid-50s to high-50s is the more forward-looking of the loyalty metrics and it moved decisively.

Geography

Q4 regional room-night growth was low double digits in both Asia and the United States, and high single digits in Europe and Rest of World. For the full year, Europe and Rest of World grew high single digits, Asia low double digits, and the United States mid-single digits. The U.S. trajectory within the year is the standout: growth improved from low single digits in the first half to low double digits in the fourth quarter, helped by targeted brand and performance-marketing investment and continued B2B momentum.

Booker mix was broadly stable, with Europe roughly half of room nights, Asia about a quarter, and the United States a low double-digit percentage.

Assessment: The U.S. inflection from low single digits to low double digits inside twelve months is the single best piece of evidence that the brand spend is working, and it directly justifies the Q4 marketing de-leverage that drew the first analyst question.

Key Topics & Management Commentary

Overall Management Tone: Assured and forward-leaning, with the emphasis shifted from defending the AI thesis to spending against it. The prepared remarks devoted more time to the supplier value proposition and to operational complexity than to demand, which reads as a deliberate reframing of the competitive debate onto ground management considers defensible. Where the commentary was least satisfying was on the return profile of the $700 million reinvestment, which was quantified for one year and not beyond.

1. Room Nights Accelerated in All Four Quarters of 2025

Q4 room nights of 285 million grew 9.0% and exceeded the high end of guidance by three percentage points, driven by continued investment in Asia and the U.S. and by a booking window that was again more expanded than expected. The quarterly progression through 2025 was 7.2%, 7.7%, 8.2% and 9.0%.

"We finished off a year of steady improvement with our room night growth accelerating each quarter throughout 2025." — Ewout Steenbergen, CFO

Assessment: This directly refutes the concern we raised at Q3, where we read the 4% to 6% guide as embedding a pull-forward payback and modelled second-half room nights blending to roughly 6.5%. The actual blend was about 8.6%. The booking-window expansion was not a one-quarter borrow; it persisted, and the guide was conservative rather than cautionary. We were wrong on that point and have removed the assumption.

2. What the 2026 Guide Actually Says

Full-year 2026 guidance calls for low double-digit reported growth in gross bookings and revenue, Adjusted EBITDA growing faster than revenue with roughly 50 basis points of margin expansion, and adjusted EPS up mid-teens. Against 2025's reported 13.4% revenue growth and 21.9% adjusted EPS growth, that reads as a deceleration, and the tape treated it as one.

The constant-currency picture is different. Management guided constant-currency top-line growth to roughly 100 basis points above the long-term ambition of at least 8%, and constant-currency adjusted EPS growth in line with the 15% ambition. Against 2025's constant-currency actuals of 10% revenue and 18% adjusted EPS, the deceleration is about one point on the top line and three on the bottom. Currency accounts for much of the rest: the FX tailwind falls from roughly 300 basis points on revenue and 400 on adjusted EPS in 2025 to about 200 and 150 respectively in 2026.

Assessment: A seven-point optical step-down in adjusted EPS growth decomposes into roughly 2.5 points of currency roll-off and roughly 3 points of deliberate reinvestment, leaving something like 1.5 points of genuine underlying moderation. The market priced the optical number.

3. The $700 Million Reinvestment

This is the central fact of the guide. Booking is reinvesting about $700 million above baseline in 2026, against roughly $170 million in 2025, into generative AI, the Connected Trip, Asia and U.S. growth, advertising, OpenTable's international expansion, fintech and loyalty. Management expects roughly $400 million of incremental revenue in 2026 from these initiatives, for a net Adjusted EBITDA impact of about $300 million.

The funding is internal. Transformation Program in-year savings are guided to $500 to $550 million in 2026, more than $250 million above 2025, and the program has already enabled approximately $550 million of annual run-rate savings, the high end of the prior target.

"So it's largely self-funded through our initiatives. So therefore, we can still expand our EBITDA margins by approximately 50 basis points and grow the top line 100 basis points faster than our overall algorithm." — Ewout Steenbergen, CFO

Assessment: A $300 million net drag on a $9,937 million Adjusted EBITDA base is about three points of growth, which is most of the gap between the 2025 result and the 2026 guide. The relevant question is not whether the drag exists but whether $700 million of spend that returns $400 million of revenue in its first year is a good use of capital. For a company whose multiple has compressed specifically because investors fear AI-driven disintermediation, declining to spend would be the harder decision to defend.

4. Marketing De-Leverage: Choice, Not Symptom

Marketing rose to 4.5% of gross bookings from 4.2%, a 25 basis point de-leverage that drew the first question of the call. Management gave three causes and defended all of them: opportunistic performance-marketing spend at attractive returns, a 13% increase in social-media investment, and back-weighted brand spend against a 2024 comparison concentrated around the summer football championship.

"So we see time where we can put money to work that's going to help build the value of this franchise over the long term, we're going to do it. And I'm not going to worry about the fact that, oh, it looks like a little deleverage here." — Glenn Fogel, CEO

The CFO added the context that matters, noting the de-leverage occurred in a quarter delivering 19% Adjusted EBITDA growth and 80 basis points of margin expansion, and committed to marketing leverage in 2026 absent further opportunistic moments.

Assessment: Defensible, and the evidence supports it: the spend coincided with room nights beating by three points and U.S. growth reaching low double digits. The full-year marketing ratio was flat. We would treat a second consecutive quarter of de-leverage without a corresponding volume beat very differently.

5. The Merchant-of-Record Argument Against Disintermediation

Asked what the language-model partners are likely to do strategically, the CEO delivered the most detailed articulation yet of why he believes they will stay at the top of the funnel. The argument runs through supply complexity (over 4 million properties, mostly unsophisticated independents requiring constant connectivity and human partner-services support), payments complexity (100-plus payment methods, 50-plus currencies), and regulatory complexity across more than 200 countries.

"So do we think that the large language models are going to be entering and want to enter down the funnel down to where we are? I don't think so." — Glenn Fogel, CEO

He then made the more interesting point: that the profitable equilibrium for a top-of-funnel player is the Google model of auctioning traffic rather than fulfilling transactions, and that the same logic likely applies to language models.

Assessment: This is a genuinely stronger version of the argument than the "it did not happen with Google" framing offered last quarter, because it identifies the specific economic reason it did not happen: intermediating transactions is a low-margin, high-complexity, heavily regulated business that a high-margin advertising platform has no incentive to enter. That is a structural argument rather than an analogy.

6. Generative AI Shows Up in the P&L

The most quantified AI disclosure again came on the cost side, where customer-service cost per booking fell roughly 10% while absolute customer-service costs declined on double-digit volume growth. The CFO named the point directly.

"I think a colleague CFO, said during our call last quarter, I've forgotten who it was, GenAI is everywhere except for in the P&L. I'm very proud that we can point to a line item in the P&L where we actually do have meaningful results on the efficiency basis." — Ewout Steenbergen, CFO

On the revenue side the disclosure stayed qualitative: more engagement, faster search, better conversion, lower cancellation rates, higher satisfaction, with the explicit caveat that "these numbers are still relatively small."

Assessment: The asymmetry persists and is now a full year old. AI is a proven, sized cost lever and an unproven, unsized revenue lever. Sales and other expenses were a source of year-over-year leverage in Q4 despite rising payments costs from the merchant mix, which is the clearest evidence yet that the cost benefit is real and compounding.

7. Take Rate Rose in 2025

Full-year revenue as a percentage of gross bookings was 14.5% against 14.3% in 2024, driven primarily by higher payments revenue, with management stating that underlying accommodation take rates continue to be stable.

Assessment: This corrects the read we carried out of Q3, where a 30 basis point quarterly decline led us to model take rate declining 20 to 30 basis points a year on flight mix and merchandising. On a full-year basis it rose 13 basis points. The quarterly figure was seasonal and mix-distorted; the annual figure is the one to model, and payments revenue is more than offsetting the dilution from flights. We have removed the declining-take-rate assumption.

8. Genius Reaches a High-50s Share of Room Nights

Genius Level 2 and 3 travellers accounted for a high-50s percentage of Booking.com room nights in 2025, up from the mid-50s in 2024, while remaining just over 30% of the active base. These travellers have a meaningfully higher direct booking rate than others. Management framed further Genius development as a 2026 priority and connected it explicitly to the agentic strategy.

"We're going to build a lot of agentic tools at the top of the funnel, so people that can come to us in our platform that they like, they trust, they have the loyalty with." — Ewout Steenbergen, CFO

Assessment: A loyalty tier representing 30% of users and nearly 60% of volume is a concentration that works strongly in Booking's favour, because it means the marginal booking increasingly comes from a customer with a stored relationship rather than from an auction. This is the most direct counterweight to the disintermediation risk in the entire disclosure set.

9. Capital Return Steps Up

Booking returned $8.2 billion to shareholders in 2025, comprising $5.9 billion of repurchases and $1.2 billion of dividends, plus $1.1 billion used to settle the convertible conversion premium in cash and avoid share issuance. Q4 alone saw $2.1 billion of repurchases and roughly $2.4 billion of total capital return, the highest quarterly figure since 2023. The dividend was raised 9.4% to $10.50 per share. Remaining repurchase authorisation stands at $21.8 billion.

The cumulative record is the striking part: since restarting repurchases in early 2022, the company has bought back $29 billion of stock and reduced the share count by 22% net of stock-compensation dilution. Management framed the capital-structure target as investment-grade ratings with roughly 2x gross leverage through the cycle.

Assessment: Returning more than 100% of free cash flow while maintaining investment-grade ratings and growing is the correct posture for a business with negative working capital and no large acquisition need. The pace stepped up materially into the share-price weakness, which is what a disciplined repurchase program is supposed to do.

10. The 25-for-1 Stock Split

The board approved a 25-for-1 forward split, effective April 2, 2026, with post-split trading beginning April 6. At the reaction-day close of $4,007.50 the post-split equivalent is $160.30.

Assessment: Economically neutral and worth noting only for two second-order effects. A four-figure share price is a practical obstacle to retail participation and to index or options mechanics, and removing it modestly broadens the buyer base. It also signals that management does not expect to need the optical scarcity of a high share price, which is a mild confidence tell at a moment when the stock is making 52-week lows.

11. Alternative Accommodations Slowed, and Management Pushed Back

Alternative accommodation room nights grew about 10% for the full year and about 9% in Q4, against a global mix of roughly 36% of room nights, up one point. The deceleration prompted a direct question about whether the category can return to double-digit growth, and whether its absence from the named 2026 investment areas signalled deprioritisation.

Management rejected that reading and supplied supporting data: supply grew roughly 8% to 8.6 million listings, and U.S. listings are growing faster than the global average.

"So definitely, you should interpret here that we are quite optimistic about the outlook for alternative accommodations." — Ewout Steenbergen, CFO

Assessment: A fair pushback, though the CEO's admission that inventory is "nowhere near where I'd like to be" is the more informative half of the answer. Alternative accommodations remain the clearest structural gap against the category leader, and 8% supply growth is adequate rather than aggressive.

Guidance & Outlook

The framing management chose is the one worth repeating: constant-currency top-line growth approximately 100 basis points ahead of the long-term algorithm of at least 8%, with constant-currency adjusted EPS growth in line with the 15% ambition. That is a company guiding above its own stated algorithm on the top line, in the same release that took the stock to a 52-week low.

Implied Q1 ramp: The Q1 Adjusted EBITDA guide of 10% to 14% looks like the weakest line in the outlook until it is normalised. Management noted the high end would be roughly 20% after adjusting for $53 million of one-time benefits in the prior-year first quarter. Q1 revenue growth of 14% to 16% against room nights of 5% to 7% implies a nine-point wedge from ADRs, vertical mix and a seven-point currency benefit, consistent with the full-year framework.

Street at: Sell-side revenue growth for the coming year was running near 8.6% ahead of the print, against a trailing three-year revenue CAGR above 16%. Management's low double-digit reported revenue guide is therefore above that consensus mark, which makes the sell-off harder to attribute to the top line and easier to attribute to the adjusted EPS line and to the margin-expansion step-down from 193 basis points to about 50.

Guidance style: Conservative, and now with a two-quarter track record we can point to. The Q4 guide of 4% to 6% room nights produced 9%, and the guide was cleared on all four metrics by three to five points. The Q1 room-night guide of 5% to 7% should be read with that pattern in mind. The one change worth noting is that management declined to guide full-year room nights at all this year, having given an "about 7%" full-year figure at this stage last year.

Analyst Q&A Highlights

Whether the Fourth-Quarter Marketing De-Leverage Marks a Strategy Change

The call opened on the marketing line, with the questioner noting the de-leverage looked like a departure from a long-established pattern and that the 2026 commentary implied more of the same. Management characterised it as opportunism funded by savings elsewhere, and committed to leverage in 2026 absent further opportunities.

Q: "The question I wanted to ask had to do with the marketing spend deleverage in the December quarter. Could you just talk about that a little bit more? Was there something that you saw in particular that made you want to lean in? And it sounds like you're going to continue to lean into marketing spend in '26. But just any more color on that? It seems like a little bit of a change for you."
— Mark Mahaney, Evercore ISI

A: "First, I think it's important to look at the marketing deleverage in the context of a quarter where we delivered 19% EBITDA growth and expanded EBITDA margins by 80 basis points… And also, please note that for 2026, what we have said is that we aim to achieve marketing leverage again unless, of course, we find those really great opportunistic moments that we can lean in again in this year."
— Ewout Steenbergen, CFO

Assessment: The right context was supplied without being asked for. Margin expanded 80 basis points in the same quarter the marketing ratio rose 25, so the spend was funded rather than absorbed. The 2026 commitment to leverage gives a checkable marker for next quarter.

Where Agentic Capabilities Produce Economic Uplift

A question on where agentic tools change consumer behaviour enough to matter economically, and on the 2026 product cadence, produced a deliberate refusal on timing and a short list of measured effects. Management declined to name specific rollout areas, describing the request as asking for the playbook.

Q: "As you roll out agentic capabilities across your properties in the coming months and years, where do you see the greatest economic uplift for bookings from these technologies in terms of how you expect them to change consumer behavior? And relatedly, what should we expect in terms of, say, the product cadence you intend to deliver in terms of getting these agentic experiences into the hands of the consumer in 2026?"
— Lee Horowitz, Deutsche Bank

A: "More engagement from our traveler customers, faster search, better conversion, lower cancellation rates and positive customer satisfaction. So those are a couple of things we're seeing. Again, the numbers are relatively small, but certainly encouraging."
— Ewout Steenbergen, CFO

Assessment: The same five qualitative effects were given last quarter with the same "relatively small" caveat. A full year into the agentic push, the revenue-side impact remains unquantified, which is the single largest disclosure gap in the story.

What the Scaling Loyalty Program Signals About Future Investment

A question on what the growing Genius base signals and whether it warrants incremental investment over the next twelve to eighteen months drew an unusually candid statement that the program is well short of its potential, plus a data point tying loyalty tier directly to direct-booking behaviour.

Q: "I want to come back to the comments you made about the Genius program. Can you talk to us a little bit about what signal you're getting as the scale of the Genius program continues to build? And how do you think about that as an area for incremental investments in the business to continue to gain operating momentum around the Genius program when you think about your priorities for the next 12 to 18 months?"
— Eric Sheridan, Goldman Sachs

A: "I think that Genius is a wonderful program, but it's not nearly what it can be and will be in the future. Look, we know how important loyalty is, particularly nowadays, the way we see the world changing. And we want to increase that loyalty."
— Glenn Fogel, CEO

Assessment: "The way we see the world changing" is management acknowledging without saying so that loyalty is the strategic answer to AI-mediated discovery. Loyalty was named as one of the 2026 reinvestment areas, which makes this the clearest link between the spend and the risk.

What the Language-Model Partners Are Likely to Do Strategically

The most substantive exchange of the call asked what behaviours and strategies the AI partners are adopting, and specifically whether they intend to become merchants of record. Management disclaimed inside knowledge and then argued from economics that they will not.

Q: "I wonder if you could give us an update on how their plans are evolving, what behaviors you're seeing, strategies that they're adopting in terms of how they sell traffic or anything they're attempting to do, potentially, if you could talk about their views on being a merchant of record, that would be hugely helpful."
— Alex Brignall, Rothschild & Co Redburn

A: "It's been very successful for that large person at the top of the funnel, Google. We've both done very well. They've made very good money setting up an auction format for taking advertising… And by the way, some of these other large language models when they're looking at this, they may say, that's the right way to do it. Not go down the funnel, don't become a merchant of record."
— Glenn Fogel, CEO

Assessment: The strongest formulation of the bull case the company has produced. It replaces an argument from precedent with an argument from incentives: advertising is a better business than fulfilment, so the rational move for a top-of-funnel platform is to auction traffic rather than to absorb the operational and regulatory burden below it.

Whether Paid Search Is Still Growing, and Where the Reinvestment Lands

A two-part question sought confirmation that traditional paid search remains a growth channel and asked which expense lines absorb the $700 million reinvestment. The answer on search was brief and affirmative; the answer on the spend supplied the funding math.

Q: "Are you still seeing growth in paid search — traditional paid search channels? And then second one would just be looking at some of the reinvestment, I think you said $700 million… But is a lot of that show up in marketing spend? Does it show up in supply acquisition people to build out experiences? Or like where — if you can elaborate a bit on where some of that spend is hitting."
— Lloyd Walmsley, Mizuho

A: "Those expenses then go into 2 lines, fixed OpEx and sales and other. But we will have also then revenues that are offsetting. So we said a net impact on the EBITDA line is about $300 million. But we also said that the Transformation Program is delivering additional in-year savings in 2026 of $250 million. So it's largely self-funded through our initiatives."
— Ewout Steenbergen, CFO

Assessment: The most useful modelling answer on the call. The spend lands in fixed operating expenses and sales and other rather than marketing, which is why the company can commit to marketing leverage and a $700 million investment simultaneously. The confirmation that paid search is still growing is a quiet but important negative data point for the disintermediation thesis.

Whether Alternative Accommodation Growth Can Re-Accelerate

The closing question flagged continued deceleration in alternative accommodations and asked whether its omission from the 2026 investment list indicated deprioritisation. Management rejected the inference and countered with supply data.

Q: "In your alternative accommodation business, continuing to slow a little bit there, albeit with very tough compares… Should this return to double-digit growth territory on a room night basis if you expect stronger top line for total overall company? And is there anything in particular you need to do to reaccelerate that growth? I don't recall hearing alts being called out as kind of one of the investment areas for this year."
— Trevor Young, Barclays

A: "There are areas where we are still nowhere near where I'd like to be in terms of inventory, but we are building it."
— Glenn Fogel, CEO

Assessment: The candour is useful and the concession is real. Supply growth of about 8% to 8.6 million listings is respectable but not the profile of a business closing a competitive gap quickly, and no growth target was offered.

What They're NOT Saying

1. Full-year room-night guidance: Only Q1 was guided. At this point last year management gave an "about 7%" full-year room-night figure alongside the quarter. Withdrawing the annual volume guide in the year they are spending $700 million to accelerate growth is a conspicuous omission.
2. Any return profile on the $700 million beyond year one: The $400 million revenue and $300 million net EBITDA figures are 2026 numbers. No payback period, no steady-state contribution, no hurdle rate, and no indication of whether the elevated spend recurs in 2027.
3. KAYAK after the impairment: A $457 million writedown one quarter ago, no update this quarter, no question asked. Whether the customer-acquisition-cost pressure that triggered it has stabilised is unknown.
4. Language-model referral volume, still: A second consecutive quarter with no absolute number, no percentage of traffic, and no economics. The qualitative list of agentic benefits was repeated verbatim in substance from the prior call, again caveated as "relatively small."
5. Why B2C direct mix stopped rising: The full-year figure was described as mid-60s and "similar to last year," where the trailing figure given at Q3 was mid-60s and explicitly up from the low 60s a year earlier. A plateau in the single metric that most directly rebuts the disintermediation thesis went unmentioned and unasked.
6. Attractions deceleration: Growth slowed from close to 90% at Q3 to nearly 80% for the year, and the vertical was cited as a growth engine without the deceleration being acknowledged.
7. The 2027 margin path: Guiding roughly 50 basis points of expansion after 193 raises the obvious question of whether 50 is the new normal or a one-year investment dip. Management framed 2026 as an investment year without saying what follows it.

Market Reaction

• Pre-print setup: BKNG closed at $4,270.00 on February 18 ahead of the after-market release, already down 20.3% year to date against a 0.5% gain for the S&P 500, down 17.0% over the trailing twelve months and 16.5% over the trailing 30 days. The pre-print 52-week closing range was $4,140.25 to $5,816.00, so the stock entered the print within 3% of its 52-week low.
• Next-day session: Shares gapped down to open at $4,088.75, traded as low as $3,871.00 and as high as $4,141.50, and closed at $4,007.50, down 6.1%. The session low broke through the prior 52-week closing low, setting a new one.
• Volume: 22.7 million shares against an 8.6 million 30-day average, or 2.7 times normal. This was a repositioning event, not a drift.
• Benchmark: The S&P 500 fell 0.3%, so essentially all of the 6.1% decline was stock-specific.

The setup matters more than the reaction. This stock had already fallen 20% in seven weeks of 2026 before printing a quarter that beat every guided metric by three to five points. The decline is therefore the continuation of a de-rating that began well before this release, and the release supplied the excuse rather than the cause.

What the market objected to is identifiable and narrow. Reported adjusted EPS growth guidance of mid-teens against 2025's 21.9% is a seven-point step-down, and guided margin expansion of about 50 basis points against 2025's 193 is a step-down of similar visual weight. Both are true. Both are also substantially explained by currency rolling off and by a $700 million reinvestment that management sized, funded and defended in the same breath. A reader who stopped at the headline guide would conclude the business is decelerating sharply; a reader who reached the constant-currency framework would conclude it is decelerating by about one point on the top line while spending against its principal strategic risk.

The gap-down open followed by a close above the intraday low, on 2.7 times normal volume, suggests forced repositioning into the print rather than a considered revaluation after it. That distinction is the basis of our upgrade.

Street Perspective

Debate: Is the 2026 guide a demand problem or an investment decision?

Bull view: The bull argument is that nothing in the guide indicates weakening demand. Constant-currency top-line growth is guided roughly 100 basis points above the company's long-term algorithm, Q1 room nights of 5% to 7% follow a Q4 that beat a 4% to 6% guide by three points, and the entire earnings step-down decomposes into currency roll-off plus a deliberate, self-funded $700 million investment.

Bear view: The bear response is that investment years have a way of extending, that management declined to guide full-year room nights for the first time, and that a company confident in its demand outlook would not need to spend $700 million above baseline to hold a 9% constant-currency top line. The $400 million first-year revenue return on $700 million of spend is also not obviously a good trade.

Our take: The bulls have this. The withheld room-night guide is a fair flag, but the substance points the other way: room nights accelerated in all four quarters of 2025, Q4 beat by three points, and paid search is still growing. The spend is best understood as buying insurance against the exact risk the multiple already discounts, and it is funded by savings rather than by margin.

Debate: Has the AI de-rating gone too far, or not far enough?

Bull view: The optimists point out that the multiple has compressed from roughly 22.5 times forward earnings four months ago to about 15.3 times now, while the fundamentals accelerated over the same period: room nights from 8.2% to 9.0%, constant-currency revenue from 8% to 11%, full-year take rate up rather than down. A stock cannot de-rate 30% on a risk while the operating evidence for that risk fails to appear.

Bear view: The skeptics argue the market is correctly pricing a terminal-value problem that will not show up in quarterly room nights until it is too late to react. The KAYAK impairment is the leading indicator, the plateau in B2C direct mix is the second, and the fact that management still refuses to quantify language-model referrals after a year of being asked is the third.

Our take: The bears identify the right metrics and draw too much from them. Two of the three are genuinely concerning and we are tracking all three. But the operating evidence is now four quarters old and moving the wrong way for the bear case, while the price has moved a long way in its favour. At 15.3 times forward earnings for a business guiding mid-teens earnings growth with a 22% share-count reduction behind it, the risk is being paid for rather than ignored.

Debate: Does the merchant-of-record moat argument hold?

Bull view: The bull case is that fulfilment is a genuinely hard, low-margin, heavily regulated business: 4 million properties mostly run by unsophisticated independents, 100-plus payment methods, 50-plus currencies, and a regulatory perimeter spanning more than 200 countries. A high-margin advertising platform has no incentive to descend into it, and the Google precedent shows what the profitable equilibrium looks like.

Bear view: The bear rejoinder is that the argument assumes AI platforms optimise for margin rather than for owning the customer relationship, that payments infrastructure is increasingly available as a service rather than as a moat, and that "hard" is not the same as "impossible" for firms with effectively unlimited capital.

Our take: The bulls are right about the near term and the bears are right that the moat is narrower than management's framing implies. Payment rails are rentable; supply relationships with 4 million independent properties and several thousand partner-services staff are not, at least not quickly. The moat is the supply side, not the payments side, and management would strengthen its own argument by saying so more plainly.

Model Update & Valuation Framework

Valuation. At the February 19 close of $4,007.50, BKNG trades at approximately 17.6 times FY2025 actual adjusted EPS of $228.06 and about 15.3 times our FY2026 estimate of roughly $262. At our initiation four months ago the stock traded at roughly 22.5 times forward earnings. The multiple has compressed about 32% while constant-currency revenue growth accelerated from 8% to 11% and room-night growth accelerated from 8.2% to 9.0%.

Fair value framework. We apply 17 to 19 times our FY2026 adjusted EPS estimate of roughly $262, producing a fair value range of approximately $4,454 to $4,978 with a midpoint near $4,716. Against $4,007.50 that implies about 11% to 24% upside, with a midpoint around 18%. The multiple range is deliberately well below where the stock traded in October, because the AI overhang is unresolved and we are not assuming it re-rates back. It requires only that the stock stops de-rating while earnings compound at mid-teens.

What would change the rating. Returning to Hold would require the direct-mix and Genius series to stall together, a second impairment among the discovery-dependent assets, or a 2026 in which the $700 million produces materially less than the promised $400 million of revenue. Moving to Underperform would require evidence of actual volume loss to AI channels, which nothing in four quarters of disclosure yet supports.

Thesis Scorecard Post-Earnings

Scored against the six pillars established at initiation.

Overall: Thesis strengthened. Two of the three bear points weakened on this quarter's evidence, the third is unchanged, and two of three bull pillars were confirmed outright. The one deterioration is inside Bull #1, where the B2C direct-mix plateau is a genuine yellow flag that we will treat as the primary thing to watch. Set against a 32% multiple compression since initiation, the risk/reward has moved decisively.

Action: Buy. The quarter that was supposed to validate the de-rating did the opposite, and the price moved as though it had confirmed it.