Booking Finally Put a Number on the AI Question, and It Was Reassuring: Downgrading to Hold on Price, Not Thesis

Results vs. Consensus

Q2 2026 Scorecard

Versus the Company's Own Guidance

The Street mark of $7.19 billion sat almost exactly inside the guided range of 4% to 6% growth, which implied $7,070 million to $7,206 million. Consensus had simply adopted the company's guidance, and the company beat it on all four lines, as it has in three of the last four quarters.

Year-Over-Year Comparisons

As in the prior quarter, the GAAP comparison is distorted by currency. Q2 2025 net income of $895 million absorbed $961 million of unrealised losses on Euro-denominated debt; Q2 2026 booked a $195 million gain on the same exposure. The 118% growth is arithmetic. Adjusted net income growth of 8.4% and adjusted EPS growth of 15% are the operative figures, and the six-point gap between them is share count.

First Half 2026

Revenue

Revenue of $7,352 million grew 8.1%, or about 7% in constant currency, and exceeded the high end of guidance by roughly two points. The relationship between gross bookings and revenue inverted this quarter: gross bookings grew 9.0% while revenue grew 8.1%, and management attributed the shortfall to the March cancellations flowing through with a lag.

"Revenue growth was lower than gross bookings growth, primarily due to elevated cancellations in March that affected second quarter revenue." — Ewout Steenbergen, CFO

This is exactly what we flagged a quarter ago. Because Booking recognises revenue at the time of travel rather than at booking, the March disruption protected Q1 revenue and penalised Q2. Take rate fell to 14.4% of gross bookings from 14.6% for the same reason.

Assessment: The recognition lag has now fully played through, which means the Q3 guide of 4% to 6% revenue growth is a cleaner read on underlying demand than either of the last two quarters. The item to watch is that constant-currency revenue growth of 7% is the first sub-8% print in our coverage, and the company's own long-term ambition is "at least 8%."

Margins

Adjusted EBITDA margin of 36.0% expanded 38 basis points, and the quality of that expansion is high. Adjusted fixed operating expenses grew just 6%, with adjusted personnel expenses up only 1%, reflecting the discretionary-spend controls and hiring recalibration management introduced last quarter. Sales and other expenses were 1.9% of gross bookings and levered despite the higher merchant mix, aided by a one-time benefit from processing fee reversals that should be treated as non-recurring.

Marketing was the offset, growing 10.8% against 9.0% gross bookings growth and taking marketing to 4.7% of gross bookings from 4.6%. Management named three causes: traffic mix, incremental paid marketing at attractive returns, and a shift of merchandising spend into performance marketing. That last item is a reclassification between lines rather than incremental spend, which softens the read.

Assessment: Cost discipline is doing real work, and it is worth noting that management delivered margin expansion while explicitly protecting the strategic investment budget. This is now the third consecutive quarter of marketing de-leverage, however, and the "we aim to achieve marketing leverage again" commitment made in February has not yet been met.

EPS and Cash Flow

Adjusted EPS of $2.54 grew 15%, ahead of Adjusted EBITDA growth of 9.3% because of a 5.5% reduction in the average diluted share count. That is the widest EPS-to-EBITDA wedge in our coverage window and it is entirely repurchase-driven.

Free cash flow of $3,643 million grew 16.1%, comfortably ahead of Adjusted EBITDA growth, with the seasonal deferred merchant bookings balance reaching $10,121 million at the June quarter-end. First-half free cash flow of $6,751 million grew 7.2%.

Assessment: The earnings algorithm is now carried disproportionately by the buyback. Adjusted net income grew 8.4% and adjusted EPS grew 15%, meaning roughly six of the fifteen points came from share count. That is a legitimate and durable source of return given the $14.5 billion remaining authorisation, but it is worth naming plainly: operating earnings growth is running in the high single digits, not the mid-teens.

Segment Performance

Revenue by Type

Merchant and Payments

Merchant gross bookings reached approximately 73% of total gross bookings, up about four points year over year, though the growth rate decelerated sharply to 14.5% from 24.3% in Q1. The mix has now risen roughly ten points in six quarters, from about 63% at the end of 2024.

Assessment: The deceleration is the arithmetic consequence of a high base and confirms the ceiling we flagged last quarter. With merchant at 73%, the remaining runway is roughly a quarter of the book, and the payments-driven tailwind to take rate has perhaps two years left. Modelling should not extend it further.

Agency

Agency revenues fell 6.9%, a wider decline than Q1's 2.3%, while agency gross bookings fell only 3.3%. The gap between the revenue decline and the bookings decline is worth noting because it runs opposite to the merchant migration story.

Assessment: A modest negative and the first quarter in which the agency revenue decline meaningfully outpaced the agency bookings decline. One quarter is not a trend, but if it repeats it would suggest commission compression rather than pure mix migration.

Advertising and Other

Growth of 8.4% represents a stabilisation after Q1's deceleration to 9.3% from 14.0%. Three quarters have now passed since the KAYAK impairment without an update, and no analyst has asked in four consecutive calls.

Assessment: Stabilising at high single digits is an acceptable outcome for the portfolio's most search-exposed asset in a quarter when management confirmed SEO pressure. The continued silence is now more conspicuous than the numbers.

Key KPIs

Assessment: The accommodation core held and everything attached to it slowed. Room nights beat, constant-currency ADRs posted their best reading in our coverage at about 2%, and the loyalty and app metrics stayed at their improved levels. The Connected Trip verticals, which were supposed to be the growth accelerant, all decelerated at once: flights to 3.7%, attractions to double digits, rental cars to negative, and alternative accommodations to 4% with the mix flat for the first time.

Geography and the Domestic/International Split

The defining decomposition of the quarter is not regional but by trip type. Domestic room nights, meaning travel within a traveller's own country, grew high single digits. International room nights "increased slightly," reflecting continued pressure on long-haul travel from reduced airline capacity and elevated fares.

By booker region, Europe and Asia each grew mid-single digits, with domestic components of high single digits and low double digits respectively. The United States grew high single digits, driven by domestic demand, with the direct channel still growing. Rest of World, which contains the Middle East, grew mid-single digits, a meaningful improvement from the low-single-digit decline in Q1 as Middle East bookers normalised.

Assessment: The domestic-versus-international split is the cleanest available evidence that this remains a supply-side aviation problem rather than a demand problem. Consumers are still travelling; they are travelling closer to home because long-haul seats are scarcer and dearer. The U.S. stepping back from low teens to high single digits ends a four-quarter acceleration streak and is the one regional datapoint that deserves genuine concern, though management's framing that it still far exceeds market growth is credible given the domestic-led composition.

Key Topics & Management Commentary

Overall Management Tone: The most forthcoming call of the four we have covered, and the difference was concentrated in one place: after three quarters of declining to size AI referral traffic, the CFO simply gave the number. The rest of the commentary was consistent with prior quarters, confident on execution and unusually specific about which parts of the deceleration are exogenous. Where management was least persuasive was on the Connected Trip verticals, where the explanation for flights was strong and the explanation for attractions and rental cars was absent.

1. LLM Referral Traffic Is Below 1% of Room Nights and Not Moving

The disclosure the market has waited four quarters for arrived as an aside in an answer about conversion rates. Traffic from large language models, on both a paid and unpaid basis combined, is running at significantly below 1% of room nights and has not changed materially.

"If we look at the traffic we're currently receiving from large language models… both on a paid and an unpaid basis 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." — Ewout Steenbergen, CFO

Two supporting details matter. The vast majority of organic language-model traffic is believed, per third-party data, to arrive at Booking directly rather than through an intermediary. And Booking is inside the OpenAI cost-per-click test group, which it frames as channel diversification.

Assessment: This is the most consequential disclosure in our coverage of this name. The bear case has rested on an unmeasured quantity, and an unmeasured quantity can be assigned any magnitude an investor prefers. A number below 1%, flat over several quarters, converts the debate from existential to incremental. It does not settle the long run, and management was careful not to claim it did. But a thesis that could not be falsified for four quarters can now be tracked.

2. SEO Pressure, Explained and Bounded

Last quarter's disclosure of "continued declines in SEO traffic" was pressed on directly, and the answer was both an admission and a useful clarification. The CEO attributed the pressure to Google's display changes, naming AI Overviews specifically.

"I do believe that some of the changes that were made in the display at Google definitely put some pressure on SEO. Putting in that AI overview probably has done it… our direct mid-60s hasn't gone down at all. In fact, total number, absolute number, obviously had to go up to maintain the same percentage share." — Glenn Fogel, CEO

The CFO then resolved the arithmetic that has puzzled us for two quarters.

"Although that the ratio of direct mix is stable, that is because that's a ratio of two positives. So we are growing the absolute [direct] channel and the paid channel. So, in absolute terms, both are going up, but the ratio, therefore, is staying stable." — Ewout Steenbergen, CFO

Assessment: This materially improves our read on the direct-mix plateau we flagged as the primary concern last quarter. A flat ratio produced by both channels growing is a different and far better fact than a flat ratio produced by stagnation. It is still not the rising direct mix the bull case originally assumed, so the flywheel is holding rather than compounding, and management's broadening of the framing to "what we're seeing across much of consumer Internet" is an accurate but convenient deflection.

3. Flights Decelerated From 28.5% to 3.7%

The most alarming line in the statistical supplement has a good explanation. Flight tickets grew 3.7% against 28.5% in Q1, but management supplied the offsetting economics.

"Our 4% growth of ticket is actually higher than the global airline industry. And if you then add 8% of ticket growth on top of it, so we had actually 12% growth — bookings growth for tickets." — Ewout Steenbergen, CFO

The cause is aviation supply: capacity reductions and higher fares stemming from the conflict, with Asia hit hardest because of its dependence on the Strait of Hormuz, and with European carriers pursuing a contrarian price-cutting strategy while the rest of the world raised fares. Management expects capacity to return gradually over the coming months.

Assessment: Unit growth of 4% with fare inflation of 8% producing roughly 12% flight gross-bookings growth is a materially better outcome than the headline suggests, and the claim of outgrowing the industry is plausible given global capacity cuts. The honest caveat is that this is the vertical management has spent two years positioning as the Connected Trip's primary customer-acquisition engine, and it will contribute far less to that engine at 4% than at 28%.

4. The Middle East Impact, Now Fading

Management updated its conflict quantification with more precision than in April. The Q1 impact of about two points came entirely from March cancellations; the Q2 impact was "a bit higher," extending through April and May; June began normalising and July continued; Q3 is expected to carry "a little bit less" impact than Q2. Direct demand from Middle East bookers is described as largely normalised, with the residual pressure being indirect and aviation-related.

"If you look at our full year guidance, we are at a high single-digit level for growth bookings and revenues and at mid-teens level for EPS at the high end, and that is still in line with our long-term algorithm and is also in line at a constant currency basis with our original guidance for the year despite the fact that in our assumptions, we have 7 out of 12 months impact of the Middle East." — Ewout Steenbergen, CFO

Assessment: The April planning assumption of impact through end-June followed by recovery has substantially held, which is a mark in management's favour on forecasting credibility. The constant-currency full-year guide matching the original February framework, while absorbing seven months of disruption, is the single strongest bull argument in the release.

5. Transformation Savings Raised a Third Time

The expected annual run-rate savings target rose to approximately $650 million from $550 million, with the incremental $100 million expected to be realised primarily in 2027. The programme has now been raised twice from its original $450 million.

"We found additional savings opportunities within existing categories, so no new categories… the additional $100 million mostly came from the procurement work stream." — Ewout Steenbergen, CFO

Assessment: A programme raised from $450 million to $650 million across six quarters, sourced from existing categories rather than new ones, is a genuine execution credential. The 2027 timing matters for the model because it lands in the year that carries our valuation anchor, and it is a real contributor to the earnings bridge rather than a rounding item.

6. AI Is Now Cash-Positive Inside the Business

Management gave the most detailed accounting yet of AI's own economics, addressing the objection that AI spending is a cost without a return.

"We are already seeing an ROI that is positive on our AI investments today as a company. So if I'm looking specifically about our AI costs, yes, they're going up. It could be token costs or license fees. It's still at a low single-digit level of our overall technology spend." — Ewout Steenbergen, CFO

The mechanism disclosed is cost-aware model routing, sending simple tasks to cheaper models and complex tasks to more expensive ones, with AI cost per merge request and technology cost per merge request both declining meaningfully. Customer service cost per booking continues to fall at a double-digit rate, and voice AI support now covers the majority of eligible inbound traveller calls.

Assessment: Booking is one of very few companies able to point to a specific P&L line where generative AI has produced a measurable, sustained cost reduction across multiple years. The disclosure that AI spend is low single digits of technology spend also bounds the downside: this is not a company that can be surprised by an AI capital-spending cycle.

7. The Connected Trip Verticals All Slowed at Once

Flight tickets grew 3.7%, rental car days fell 6.5%, attraction tickets grew double digits after roughly 80% for full-year 2025, and Connected Trip transactions grew low double digits after high 20s for full-year 2025. Alternative accommodation room nights grew 4% and the mix held at roughly 37% rather than expanding.

Management addressed flights thoroughly and rental cars not at all. On alternative accommodations, the CFO cited brand and regional mix, noting stronger growth from Agoda and Priceline and in the U.S. where the alternative accommodation offering is relatively smaller. The CEO was blunt about the gap.

"That 4% growth in alternative accommodations, well, I'd like that to be a lot higher, specifically want to be a lot higher in the U.S." — Glenn Fogel, CEO

Assessment: The strategy's supporting verticals decelerating simultaneously is the most substantive negative in the print, and only one of the four has a satisfying exogenous explanation. Connected Trip transactions still growing at more than twice Booking.com's total transaction growth means the strategic logic holds directionally. But at low double digits rather than high 20s, the incremental frequency and retention benefit the thesis assumes is arriving more slowly.

8. Record Capital Return, Executed Well Below the Current Price

Booking repurchased $3.7 billion excluding excise taxes and returned $4.1 billion in total, another record quarter. First-half repurchases of $7.4 billion were executed at an average price of approximately $173, against the $207.02 reaction-day close. Remaining authorisation stands at $14.5 billion. The share count fell 5.5% year over year, an acceleration from roughly 4% through 2025.

Assessment: Buying $7.4 billion of stock at an average price roughly 17% below where it now trades is exactly what a repurchase programme is supposed to accomplish and rarely does. It is also the reason adjusted EPS grew 15% while adjusted net income grew 8.4%, and investors should hold both facts together: the capital allocation is excellent, and it is carrying more of the earnings algorithm than it used to.

9. The U.S. Streak Ends

U.S. room-night growth stepped back to high single digits from low teens, ending four consecutive quarters of acceleration. Growth was domestic-led, and the direct channel continued to grow. The CEO acknowledged the step-down directly while defending the relative performance.

"We come down to high single digit, and that's less than some of the other ones last quarter, but it's still really, really good numbers compared to how the market overall was doing." — Glenn Fogel, CEO

Assessment: Given that U.S. growth was driven by domestic demand and the quarter's weakness was concentrated in long-haul international, a step-back from low teens to high single digits is a reasonable outcome rather than a failure. It does remove the cleanest momentum story in the equity, and it is the single reason we would not describe the fundamental picture as improving.

10. FX Turns Into a Headwind

Currency contributed only about one point to Q2 reported revenue growth, down from six points in Q1. For Q3, management guides FX to a roughly one-point negative impact on gross bookings and revenue, the first negative currency quarter in our coverage window, with the euro assumed at $1.15.

Assessment: The consequence is that reported and constant-currency growth converge, and then invert. Q3 guidance of 4% to 6% reported revenue growth corresponds to 5% to 7% constant currency. Investors who have grown used to reported growth flattering the underlying number now need to reverse that habit, and the optics of reported growth will understate the business for the next several quarters.

11. B2B Consolidation

Booking is merging the separate B2B units at Agoda, Priceline and Booking.com into a single platform, led by Agoda chief executive Omri Morgenshtern. The stated aim is a best-in-class combined offering for large partners such as banks, airlines and travel management companies. The CFO added an important discipline.

"It's very important to be focused in the B2B business on incrementality because it's easy to some extent, to grow a B2B business very fast, but cannibalizing your existing B2C business where the unit economics are much higher." — Ewout Steenbergen, CFO

Assessment: B2B has been credited in passing for U.S. growth across several quarters without ever being sized. Consolidating it under a single leader is sensible, and the explicit warning about cannibalisation is the kind of candour that raises rather than lowers confidence. This is a 2027 story at the earliest.

Guidance & Outlook

The full-year guide is essentially reaffirmed. Revenue, constant-currency revenue and adjusted EPS are unchanged from April, and Adjusted EBITDA is broadly unchanged. Only gross bookings was trimmed at the top of the range, and management was specific about why.

Implied Q3 ramp: Room nights of 3% to 5% with gross bookings and revenue each at 4% to 6% implies a wedge of roughly one to two points from ADRs and vertical mix, narrower than the three-point wedge in Q2. Since constant-currency guidance of 5% to 7% sits above reported guidance for the first time, the underlying assumption is constant-currency growth holding around Q2's 7%, consistent with management's expectation of a smaller conflict impact.

Street at: The Q2 consensus of $7.19 billion sat inside the guided range, meaning the Street has been marking to company guidance rather than forming an independent view. That pattern makes the Q3 guide of 4% to 6% the effective consensus, and the relevant question is whether the high-end beat repeats.

Guidance style: Consistently conservative and now with a track record of clearing the high end on every metric in three of the last four quarters. The one change is that the gross bookings cut was attributed to flights alone, with accommodations described as largely unchanged, which is a narrower and more credible attribution than a blanket reduction would have been.

Analyst Q&A Highlights

Whether AI Has Produced Measurable Gains in Conversion and Match Rates

A question seeking quantitative evidence of AI's revenue impact after roughly eighteen months of deployment drew a refusal on percentages from the CEO, then an unprompted disclosure from the CFO that answered a different and far more important question.

Q: "I would think at some level, AI can potentially materially improve personalization, which should show up in things like book-to-look ratios, match rates. So what do you — is it too early? Or have you seen any quantitative evidence that in this year, 1.5 years you've had with Penny that it's somehow improved conversion rates or increased booking per customer?"
— Mark Mahaney, Evercore ISI

A: "If we look at the traffic we're currently receiving from large language models… both on a paid and an unpaid basis 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."
— Ewout Steenbergen, CFO

Assessment: The question about conversion went unanswered for a fourth quarter. What came instead was the disclosure that has been withheld for a year, and it was volunteered rather than extracted. Management chose this quarter to put a number on the risk, which is itself informative about their confidence in it.

Whether Booking Is Losing Visibility Inside Language-Model Results

A follow-on question drew on survey evidence suggesting language models favour independent and inspirational content, and asked how Booking positions its data inside those systems. The answer conceded there is no playbook and surfaced an interesting asymmetry.

Q: "We've seen some evidence on our surveys of the LLMs raising kind of a broader mix of independent content, and they rely more on kind of inspirational content as well as kind of just standardized booking style content. Could you talk about what — obviously, it's very low volumes, but could you talk about how you — what you have seen on that side, how you're presenting the data within the LLMs?"
— Alex Brignall, Rothschild & Co Redburn

A: "Third-party people are saying, 'Hey, we're showing up an awful lot,' which is great. On the other hand, he also mentioned we're showing up a lot, but the referrals aren't showing much. So that's an interesting difference between the two."
— Glenn Fogel, CEO

Assessment: High visibility with low referral conversion is a genuinely ambiguous signal. The optimistic reading is brand reinforcement without channel dependency. The pessimistic reading is that language models are answering the question and keeping the user. Management chose the optimistic reading and did not defend it with data.

Why the Non-Accommodation Verticals Diverged From Room Nights

The sharpest challenge of the call put the vertical deceleration directly against the Connected Trip narrative. Management defended flights with detailed aviation economics and left rental cars unaddressed.

Q: "Connected Trip is obviously a big focus, and you had some positive comments there. But it looks like air really decelerated and rental cars were down a little bit. Any thoughts on why those are going to kind of a little different direction than room nights and what that means for connected trip?"
— Justin Post, Bank of America

A: "Our flights business continues to outperform the market even in a challenging overall airline environment… And please keep in mind that the biggest part of our economics is actually coming from accommodation bookings, and that wasn't so much impacted by those dynamics with respect to the airline industry."
— Ewout Steenbergen, CFO

Assessment: The flights answer is strong and quantified, converting a 4% unit number into roughly 12% bookings growth. The rental car decline of 6.5% was raised in the question and never addressed, in a quarter when it worsened from the prior quarter's 4.9% decline. That is the clearest unanswered question of the call.

Where the Middle East Impact Ended Up and What Is Assumed Next

A direct request to mark the April guidance assumption to actuals produced a month-by-month account and the most useful forward statement in the release.

Q: "Last quarter, you did quantify — I know it's difficult to quantify, but you said you're assuming a 300 basis points headwind for the second quarter. Could you just tell us where that ended up and what you're assuming for the third quarter there?"
— Kevin Kopelman, TD Cowen

A: "If you look at the second quarter, the impact was actually a bit higher than in the first quarter because we saw the impact continuing in April and in May. But as we said then in June, actually, the situation started to normalize more, and that continued in the month of July. If we look now forward to the third quarter, we expect that there will be a little bit less of an impact than we have seen in the second quarter."
— Ewout Steenbergen, CFO

Assessment: Management held itself to the prior quarter's stated assumption and reported against it, which is the behaviour an analyst should reward. The April planning assumption of disruption through June followed by normalisation substantially held, and July continuing to normalise is the most current datapoint available.

Whether SEO Pressure Reflects AI Search or a Shift Toward Paid

A pointed question asked whether the SEO decline is AI substitution or simply reduced organic visibility in favour of paid placement, and whether travellers might be going directly to hotels. The answer produced the arithmetic clarification that resolves the direct-mix puzzle.

Q: "I don't think you all have called out SEO pressure before. And so the question there is sort of do you think this is a shift to AI search or more just a function of pressure on sort of organic visibility in favor of SEM? And the follow-up would just be, it sounds like AI search traffic is still miniscule… do you think that there are higher share going directly to hotels coming out of AI?"
— Lloyd Walmsley, Mizuho

A: "Although that the ratio of direct mix is stable, that is because that's a ratio of two positives. So we are growing the absolute [direct] channel and the paid channel. So, in absolute terms, both are going up, but the ratio, therefore, is staying stable."
— Ewout Steenbergen, CFO

Assessment: The single most useful clarification of the quarter for anyone who read last quarter's direct-mix plateau as deterioration. It was not. Both channels are growing; the ratio is stable because they are growing at similar rates. The CEO's admission that Google's AI Overviews are the likely cause of the SEO pressure is also the most specific attribution the company has offered.

Whether AI Has Accelerated Internal Product Development

A question about whether AI has raised the pace of innovation drew both an enthusiastic answer and, more usefully, a disclosure of how the company polices its own AI spending.

Q: "We talk to a lot of companies, they're talking about how greater usage of AI, whether it's coding or product development, seems to have accelerated the pace of innovation for folks that have kind of jumped into the pool. So I'm just wondering at Booking, like what you're seeing in terms of your pace of product development."
— Stephen Ju, UBS

A: "The most important one from my perspective is that we are having a cost aware model routing so that we make sure that for simple task, we are using cheaper models and that for more complex tasks, we're using more expensive models. So we're measuring that, for example, by looking at our AI cost over merge request and our technology cost over merge request. And actually, those are coming down in a meaningful way."
— Ewout Steenbergen, CFO

Assessment: A level of operational specificity on AI cost management that few companies offer. The CEO's stated concern about people going "bunkers with spending money on tokens" and the CFO's per-merge-request metrics indicate this is governed rather than aspirational, which is why AI spend sits at low single digits of technology spend.

What the B2B Consolidation Is Meant to Achieve

The closing exchanges covered the merger of three separate B2B units into one platform, and what kind of customer it targets. Management framed it as a competitiveness upgrade with an explicit guardrail.

Q: "Could you talk about as you consolidate your B2B operations, what opportunities to invest might emerge in B2B? And how do you view the opportunity, whether it be share gain or just absolute growth?"
— Kenneth Gawrelski, Wells Fargo

A: "By improving the platform, making it best-in-class, we will be more competitive than we already are… I am really looking forward to the future when you'll be saying, 'Gee, wow, that B2B business is really going great,' and I expect that to happen."
— Glenn Fogel, CEO

Assessment: B2B has been credited for U.S. momentum across several quarters without ever being sized, and this is the first structural disclosure about it. The CFO's caution about cannibalising higher-margin B2C is the more valuable half of the answer, and it argues for modelling B2B as a mix headwind to take rate rather than a clean growth addition.

What They're NOT Saying

1. Why rental car days fell 6.5%: Raised explicitly in a question, answered only for flights. Two consecutive quarters of decline, widening from 4.9% to 6.5%, with no explanation offered in either.
2. Why attractions decelerated from roughly 80% to double digits: A vertical repeatedly named as a growth engine, now slowed for two consecutive quarters, and discussed in neither.
3. Any quantitative evidence of AI improving conversion: Asked for a fourth consecutive quarter, declined a fourth consecutive time. The CEO explicitly said "we're not at a stage where we're going to start giving away percentages, numbers, increases in CSAT."
4. KAYAK, three quarters after the impairment: Still no update, and no analyst has asked in four consecutive calls, in a period when management has confirmed SEO pressure that is directly relevant to that asset.
5. Full-year room-night guidance, for a third consecutive quarter: Withheld in February, April and now August. The conflict explains two of the three.
6. The size of the B2B business: Consolidated under a named leader, credited for U.S. momentum, and never sized in revenue, bookings or margin terms.
7. What marketing leverage looks like in practice: Committed to in February and missed in both quarters since. Marketing as a percentage of gross bookings has risen year over year in each.
8. Whether the elevated buyback pace is sustainable: $7.4 billion in six months against $14.5 billion remaining leaves roughly one year at the current rate, and no framework was offered for what happens after.

Market Reaction

• Pre-print setup: BKNG closed at $194.27 on August 4 ahead of the after-market release, down 9.3% year to date against a 13.0% gain for the S&P 500, and down 11.3% over the trailing twelve months, but up 5.3% over the trailing 30 days. The pre-print 52-week closing range was $154.13 to $228.83.
• Reaction session: Shares gapped up to open at $207.74, a 6.9% advance, traded between $202.13 and $211.47, and closed at $207.02, up 6.6%. The stock held nearly all of its opening gain through the session.
• Volume: 10.7 million shares against a 7.1 million 30-day average, or 1.5 times normal.
• Benchmark: The S&P 500 fell 0.2%, so the entire move was stock-specific.

This is the mirror image of February. Then, a quarter that beat every guided metric was sold 6.1% to a new 52-week low. Now, a quarter with visibly decelerating volume growth was bought 6.6%. In both cases the news was secondary to what was already priced.

Three things drove the buying. The clearing of all four guided metrics, for the third time in four quarters, established a pattern rather than an event. The reaffirmation of full-year constant-currency guidance in line with the original February framework, while absorbing seven months of assumed conflict impact, removed the fear that April's cut was the first of several. And the disclosure that language-model referral traffic sits below 1% of room nights and is not moving retired the question that had been the principal justification for the multiple compression since late 2025.

The pattern of the session, a gap up that was held rather than faded, is the opposite of October's gap up that faded and April's gap down that was recovered. It reads as re-rating rather than repositioning. The stock is now up 29.1% from where we upgraded it in February and up 19.0% since our last report, and it has closed most of the valuation gap that made the upgrade compelling.

Street Perspective

Debate: Does the sub-1% LLM disclosure settle the disintermediation question?

Bull view: The optimists argue the number is decisive for any reasonable investment horizon. Referral traffic below 1% of room nights, flat over several quarters, with the majority of organic language-model traffic arriving directly, means the substitution the bears have modelled is not happening. Meanwhile the platforms themselves are choosing advertising over fulfilment, which is the structure that made Google a partner.

Bear view: The skeptics respond that a low base is exactly what an S-curve looks like before it inflects, that management disclosed the number only once it was comfortable, and that the high-visibility, low-referral asymmetry the CEO described is consistent with language models answering questions rather than passing users along. SEO pressure from AI Overviews is already visible in the funnel.

Our take: The bulls have won this round decisively and should not be allowed to claim more than they have. A flat sub-1% share over several quarters is genuine evidence and it removes the case for a large permanent discount. It says nothing about the trajectory two or three years out, and the honest position is that the risk has moved from unquantifiable to merely uncertain. That is worth several multiple points, which is roughly what the stock has just paid itself.

Debate: Is the vertical deceleration exogenous or structural?

Bull view: The bull case is that this is an aviation supply shock, not a demand problem. Flight unit growth of 4% against roughly 8% fare inflation still produced about 12% flight gross-bookings growth and outpaced a shrinking global airline market. Domestic room nights grew high single digits while international grew only slightly, which is exactly the signature of constrained long-haul capacity. Management expects capacity to return over the coming months.

Bear view: The bears note that the conflict explains flights and possibly attractions, but not rental cars declining for two consecutive quarters and worsening, and not alternative accommodations growing 4% with the mix flat for the first time. Connected Trip transaction growth has fallen from high 20s to low double digits in three quarters, which is a large move for a strategy pitched as the primary growth accelerant.

Our take: Split, and the split matters. The flights explanation is well-evidenced and we accept it. The rental car and alternative accommodation deceleration is not adequately explained and management declined to address the former when asked directly. We are treating roughly two thirds of the vertical slowdown as exogenous and one third as unexplained, and the unexplained third is why the Connected Trip pillar is no longer scored as on track.

Debate: Has the stock priced the recovery?

Bull view: The optimists argue that roughly 20 times forward earnings is unremarkable for a business with 36% Adjusted EBITDA margins, negative working capital, a share count shrinking 5.5% a year, a cost programme raised twice to $650 million, and a resolved AI overhang. On depressed 2026 earnings the multiple overstates the true valuation, and 2027 normalises it.

Bear view: The bears point out that the multiple has expanded roughly 31% since February while constant-currency revenue growth fell from 11% to 7%, below the company's own long-term floor, and that adjusted EPS growth of 15% rests on 8.4% adjusted net income growth plus a buyback. Paying a re-rated multiple for a de-rating growth algorithm is the classic mistake.

Our take: The bears have the better arithmetic and the bulls have the better business. Both can be true, and when they are, the answer is a Hold rather than a view on direction. The specific issue is that our February upgrade was explicitly a valuation call on a 15.3 times multiple, and that argument no longer exists at 20.1 times. We would need either a return toward the high teens or evidence that constant-currency growth is re-accelerating above 8% to make the case again.

Model Update & Valuation Framework

Valuation. At the August 5 close of $207.02, BKNG trades at approximately 22.7 times FY2025 actual adjusted EPS of $9.12, about 20.1 times our FY2026 estimate of roughly $10.31, and about 17.0 times our FY2027 estimate of roughly $12.15. At our February upgrade the stock traded at roughly 15.3 times forward earnings. The multiple has expanded about 31% while constant-currency revenue growth has fallen from 11% to 7%.

Fair value framework. We continue to anchor on FY2027, since 2026 carries a management-quantified conflict drag. Applying 17 to 19 times our FY2027 estimate of roughly $12.15 gives a fair value range of approximately $207 to $231, with a midpoint near $219. Against $207.02 that implies about 0% to 12% upside, with a midpoint around 6%. The low end of our range is now the market price, which is the arithmetic definition of a stock that has caught up to its fundamentals.

Why the multiple range did not rise despite better AI news. The sub-1% disclosure genuinely reduces the risk premium and argues for a higher multiple. It is offset by constant-currency growth printing below the company's own 8% floor for the first time, and by the earnings algorithm leaning harder on the buyback. Those cancel out, so we hold 17 to 19 times rather than raising it.

What would change the rating. Returning to Outperform requires either a pullback toward the high teens times FY2027, or evidence that constant-currency revenue growth is back above 8% with the Connected Trip verticals re-accelerating. We would move to Underperform only on evidence that the LLM referral share has begun to inflect, that the direct channel is shrinking in absolute terms, or that the accommodation core has joined the verticals in decelerating.

Thesis Scorecard Post-Earnings

Scored against the six pillars established at initiation and carried through three quarters.

Overall: Thesis strengthened, valuation exhausted. Two bear points weakened this quarter, including the most important one, and Bull #1 was rehabilitated by a clarification we had been waiting two quarters for. Against that, Bull #2 broke and Bear #3 shifted from a currency problem we expected to resolve into a share-count dependency we did not. The net is a better business than we thought on the risk that mattered most, growing more slowly than we assumed on the verticals that were supposed to accelerate it, at a price 29% higher than when we upgraded.

Action: Take profits on the trading position, hold the core. This is a downgrade on price rather than on thesis, and we would be buyers again toward the high teens times FY2027 earnings.