Expedia is a competent but second-best online travel agency whose owner earnings have recovered under a multi-year platform rebuild, yet its single largest customer-acquisition channel — Google search — is being rewired by AI Overviews and zero-click answers. At $251.84 vs. base IV $273.85, there is no margin of safety for a business whose moat we cannot honestly call wide.

Plain English

Expedia is the second-biggest website where you book hotels and flights. It owns Expedia, Hotels.com, and Vrbo. It makes money by taking a cut of every booking. Two big problems: a competitor called Booking.com is bigger and more profitable, and Google (where most travelers start their search) is starting to answer travel questions directly inside its results page, which means fewer people click through to Expedia. Expedia is well run today and gives cash back to shareholders, but its long-term position depends on things outside its control. The stock is fairly priced, not cheap.

Thesis

Expedia Group is the world's #2 pure-play online travel agency, behind Booking Holdings, plus a leading vacation-rental platform (Vrbo) competing with Airbnb and a majority stake in metasearch site trivago. The 2025 10-K reports $14.7B of revenue split 70% merchant / 22% agency / 8% advertising, with three core consumer brands (Expedia, Hotels.com, Vrbo) now sharing one front-end stack and one loyalty program (One Key). Owner earnings TTM are $1.67B and the company converts ~87% of net income to FCF (fcf_conversion_5y = 86.7%).

The compounder case is a four-step argument: (1) the platform consolidation under prior CEO Peter Kern is finally yielding margin and a single test-and-learn surface; (2) B2B (powering airline / fintech / GBT travel) is growing faster than B2C and is structurally less marketing-dependent; (3) capital allocation is now overtly shareholder-friendly — Q3 2025 alone returned $451M in buybacks and a reinstated $0.40/quarter dividend on a clean balance sheet (net debt / EBITDA = 0.20); (4) at 28x trailing earnings but only 15x EV/FCF, the market is pricing modest forward growth — reverse-DCF implies just 4.7% perpetual growth, well below the 6-8% travel TAM.

The price/IV math, however, is uninspiring. Composite score is 66/100, propped up almost entirely by capital allocation (20) and valuation (17) while profitability (15) reflects a 10-year average ROIC of just 1.31%. Base IV is $273.85, low IV $157.50, high IV $413.16; px/IV = 0.92. To buy with a Buffett-grade margin of safety I want the price closer to the low IV — call it the high $170s. Today the stock is interesting but not cheap.

Moat

Expedia presents a moat story that is real but visibly narrower than the bull case requires. I work through the five moat types in order.

1. Pricing power. Limited. EXPE's take rate is set by a global hotel-supply competition with Booking.com, Airbnb, Google Hotels, and the chains' own direct-booking initiatives (Marriott Bonvoy, Hilton Honors). Hotels can — and do — push direct bookings, and chain brands.com sites now use rate-parity workarounds, OTA-rate matching, and member-only pricing to pull demand off Expedia. The 10-K itself flags that 'our long-term success and profitability depend on our continued ability to maintain and increase the overall number of traveler transactions flowing through our brand and shared platforms in a cost-effective manner.' That is the language of a price-taker, not a price-maker. Verdict: weak.

2. Switching costs. Near zero on the consumer side. A traveler can shop Expedia, Booking, Google, Kayak, and the hotel's own site in five tabs. One Key (the unified loyalty program, fully rolled out in 2025) is the only meaningful switching-cost lever, and it is structurally inferior to a hotel chain's points program because the redemption value is not differentiated. On the supply side, switching costs are higher — a hotel that integrates with Expedia's distribution stack, revenue-management feeds, and ETP toggles (Expedia Collect vs. Hotel Collect) is sticky — but supply is multi-homed: virtually every property is also on Booking. Switching costs on the B2B side (Rapid API powering airline retailing, white-label sites, GBT's 10-year supply agreement struck when Egencia was sold) are the strongest in the company. Verdict: narrow on supply/B2B, none on consumer.

3. Network effects. Two-sided marketplace logic exists — more properties attract more travelers, who attract more properties — but Booking has the same network at greater scale (~28M reported listings vs. EXPE's 3.6M lodging properties, including 2.4M Vrbo alternative accommodations). Network effects accrue to #1, not #2, when consumers can costlessly visit both. Vrbo had its own US-centric whole-home-rental network effect, but the 2020-2023 platform migration onto Brand Expedia disrupted host relationships and product specificity, costing share to Airbnb that the company has not fully recaptured.

4. Intangibles (brands). The Expedia, Hotels.com, and Vrbo brand-awareness moats are the asset Buffett would most likely respect — these are decades-old top-of-mind names, and even in the AI-search era a traveler who types 'hotels.com' directly bypasses Google ranking. But Expedia spends heavily to maintain that direct demand: selling and marketing is the single largest expense line in every quarter, and a meaningful fraction of marketing is paid to Google itself. Buffett's GEICO analogy is illustrative by inversion: GEICO's brand is reinforced by a real underlying cost advantage that funds the gecko, and competitors 'are unable to cross' that moat [2][3]. Expedia's brand is reinforced by paid marketing — competitors absolutely can cross it by paying Google more. Verdict: narrow brand intangible, with rented reinforcement.

5. Cost advantages. This is where the platform consolidation matters. By collapsing Hotels.com and Vrbo onto the Brand Expedia front-end stack in 2023 and unifying loyalty/marketing tech, Expedia has lower per-incremental-booking technology cost than smaller OTAs and a structurally better marketing-mix optimization than running three siloed brands. But Booking has the same scale advantage and operates with materially higher EBITDA margins (~32% LTM vs. EXPE in the high teens to low 20s). Being the second-lowest-cost producer is not a moat — it is a vulnerability if the lowest-cost producer ever decides to compete on price.

The Buffett canon is unambiguous on what a real moat looks like: GEICO's cost advantage 'is the factor that has enabled the company to gobble up market share year after year. Its low costs create a moat — an enduring one — that competitors are unable to cross' [2]. EXPE has the structure, but not the gobble-up dynamic; share has been roughly flat-to-down vs. Booking and Airbnb for a decade. The 1996 letter notes GEICO's strategy was to widen the price advantage offered to customers, not the profit margin [3]. EXPE's One Key and member pricing gestures in that direction, but on a smaller scale and against tougher direct competitors.

Moat verdict: NARROW.

Management

Expedia's capital-allocation report card has improved sharply since the 2020 reset. I score it across the five canonical levers plus communication.

1. Reinvestment in the business. Most of what the prior CEO Peter Kern (2020-May 2024) did was internal: shut down Egencia, exited businesses, consolidated three front-end stacks into one, unified loyalty and marketing tech. These were value-creating reinvestment decisions because they reduced complexity rather than chased growth. New CEO Ariane Gorin (since May 2024, ex-head of B2B) has continued the same playbook with appropriate continuity. The pivot to AI inside the product (developer productivity, customer-service resolution speed) is reasonable but unproven. Maintenance capex is uncertain enough that the scorer flagged it — '>50% spread' — and widened the IV range. That uncertainty is real and is the strongest argument against high conviction.

2. Acquisitions. Expedia's acquisition history is mixed. trivago (2013) is now an underperforming majority stake whose advertising revenue is structurally tied to the same Google traffic that threatens its own parent. Vrbo (HomeAway, 2015, ~$3.9B) is a long-tailed problem: the company is still rebuilding share lost during platform migration. Orbitz (2015, ~$1.6B) was sensible cost takeout. Net: the M&A track record is one I would not extrapolate. The good news is that current management has explicitly stopped doing large deals — the last several years have been divestitures, not acquisitions. Buffett-style restraint on M&A counts in management's favor [1].

3. Debt. Net debt / EBITDA is 0.20 — essentially clean. The balance sheet is no longer a constraint or a risk. Interest coverage is undefined in the scorecard because interest expense is small relative to operating cash flow. A+ here.

4. Buybacks. This is the biggest story. Q3 2025 alone repurchased 2.3M shares for $451M (avg ~$196). Trailing share count is down ~1.25% over 10 years (modest, because heavy 2017-2019 issuance offset post-COVID buybacks), but the more relevant signal is the pace and the price: management has been buying in the $150-$200 range, comfortably below my base IV of $273.85, which is the textbook Buffett standard for value-creating repurchases. The scorer flagged 'net capital return period; ROIIC not meaningful' — translation: the company is now a cash-return story, not a reinvestment story. That is appropriate for a #2 in a mature category. Grade on buybacks alone: A-.

5. Dividends. Expedia reinstated a dividend in 2024 and is now paying $0.40/quarter ($1.60/year, ~0.6% yield). Symbolic more than economic, but it signals confidence in cash-generation durability and broadens the shareholder base. Neutral.

Communication and incentives. Disclosure quality is solid (dual segments B2C/B2B/trivago, gross bookings disclosed, ETP mix discussed). The Liberty/IAC/Diller historical overhang on the Class B share structure remains; Diller stepped back as Chairman in 2024. The dual-class structure is a minor governance demerit but not a deal-breaker — it is similar in spirit to many media companies and Berkshire itself. Executive compensation is heavily stock-based ($113M SBC in Q3 2025 alone, annualizing ~$450M against $1.67B owner earnings — call it 25% of OE), which is high. This is the discount that comes with being a tech-adjacent travel platform, but it materially reduces real per-share owner earnings and is the single biggest reason I am not more enthusiastic about the buybacks (a chunk of repurchase activity is offsetting dilution rather than shrinking the share count).

Synthesis. Management has done the hard, unglamorous work of integrating an unwieldy collection of brands and is now returning cash at attractive prices. They are also lucid about what they are — a #2 in OTA, a leader in B2B distribution, a platform business — rather than pretending to be the next Booking. The two material demerits are SBC creep and the inherited M&A graveyard.

Capital allocator: B+

Industry

Online travel distribution is a structurally OK business at the industry level and a structurally hard one for the #2 player.

1. Threat of new entrants — HIGH and rising. Capital is no longer the entry barrier; supply is. But the supply side is increasingly being reaggregated by Google (Hotels, Flights, Vacation Rentals, all served inside the SERP), by Airbnb (which has expanded into hotels and experiences), and now by AI assistants (ChatGPT Operator, Perplexity, Anthropic's Computer Use, and Google Gemini) that can call APIs directly. Each of those entrants partially disintermediates the traditional OTA UX. The 10-K explicitly identifies 'internet search, metasearch and social and digital media' as required marketing channels — Expedia is paying the toll to its largest competitor (Google).

2. Bargaining power of suppliers — MEDIUM-HIGH. The major hotel chains (Marriott, Hilton, Hyatt, IHG, Accor) have spent a decade building direct-booking incentives and member-rate programs. They tolerate OTAs as marginal-demand fillers but actively try to convert OTA-acquired guests into direct loyalty members. For independent hotels and alternative accommodations, the OTAs hold more leverage — these properties cannot afford comparable marketing budgets — but this is the lower-margin part of the business and the part most exposed to Google taking it directly.

3. Bargaining power of buyers (travelers) — HIGH. Switching cost is a single tab. Price comparison is frictionless. Loyalty programs (One Key) are a partial offset but cannot replicate hotel-chain points value. Travelers shop multi-platform by default. Customer-acquisition cost is therefore high and rising, and direct-app penetration is the only structural defense — which Expedia is investing in.

4. Threat of substitutes — HIGH. Substitutes include: hotel direct (chain.com), Airbnb (vacation rentals + a growing hotel inventory), Google Hotels / Travel (zero-click in many cases), traditional travel agents (resurgent in luxury), corporate travel programs (CWT, Amex GBT, BCD), and now agentic AI bookers. Each takes a slice of the demand pool.

5. Industry rivalry — INTENSE. Booking Holdings is the structural #1, with higher margins, a larger Europe/RoW base, and a more disciplined marketing ROI. Airbnb owns the alternative-accommodation mind-share that Vrbo once monopolized in the US. Trip.com dominates Asia. Within its weight class Expedia is winning the platform-rebuild battle, but no one in this category is taking durable share from Booking, and Booking's lead in EBITDA dollars funds a marketing budget Expedia cannot match.

Value pool location and trajectory. The OTA value pool is being squeezed at three pinch points simultaneously: Google captures more of the search funnel via AI Overviews and zero-click answers; chain.com captures more of the loyalty-eligible high-value traveler; and a long tail of vertical-specific tools (HotelTonight for last-minute, Going for cheap-flight alerts, Kayak/trivago for compare) captures specific use cases. The TAM is enormous (Phocuswright pegs global travel spend at >$2T in 2026 per the 10-K) and online penetration is still rising, so the absolute pool grows — but the OTA share of that pool is at best stable, more likely shrinking 50-100bps/year.

B2B distribution (Rapid API, white-label, TAAP, the GBT supply deal) is the bright spot. It is a less commoditized, higher-retention business with real switching costs — once a fintech or airline has built its travel-booking flow on Expedia's API, ripping it out is expensive. EXPE has not separately disclosed B2B economics enough for me to underwrite it as the durable compounder kernel, but it is plausibly that.

Industry Verdict: Average.

Inversion

I am now a short-seller. My job is not to hedge.

1. The single event that kills this. Google AI Overviews, plus the parallel AI-assistant transition (ChatGPT, Gemini, Perplexity, Apple Intelligence travel features), structurally compresses paid-search ROAS for OTAs over the next three to five years. The mechanism is concrete: when a user types 'best 4-star hotels in Barcelona for a family of four in July,' Google's AI Overview answers inline with hotel cards, photos, prices, and a 'book' button that routes to Google Hotels — not to Expedia.com. Even when the user does click out, fewer click-outs means a smaller paid-auction pool, which means Booking (with deeper pockets and higher LTV per direct user) wins more of the remaining auctions and pays more per click. Expedia's marketing line is its single largest cost; if marketing efficiency degrades 200-400bps over three years, the platform-consolidation savings are erased and EBITDA margins compress, not expand. The kill shot is not a single quarter — it is the slow, irreversible re-rating that happens when the market accepts that EXPE has become a melting ice cube on the consumer side and has to be valued as a B2B-only business at a much lower revenue base.

2. Why the moat is narrower than bulls think. Bulls point to brand awareness, a 30-year operating history, $14.7B of revenue, and 3.6M lodging properties. None of those is a moat in the Buffett sense. Brand awareness is rented from Google — the moment Expedia stops paying Google, direct-type-in traffic alone cannot sustain the booking volume needed to amortize fixed costs. The 30-year history is shared with Travelocity, Orbitz, and Hotels.com — all of which were once independent leaders and all of which Expedia bought for cents on the dollar after they failed to maintain category leadership; the same fate is available to Expedia itself. The 3.6M properties number is misleading because Booking, Airbnb, and Google all have functionally complete inventory; supply has commoditized. The real moat — B2B distribution — is real but is 20-25% of the business, not the whole company.

3. Why management is worse than it appears. Capital allocation looks great in a screenshot but contains an ugly footnote: stock-based compensation runs ~$450M annualized against $1.67B owner earnings, meaning roughly a quarter of nominal owner earnings is being recycled to employees, and a meaningful fraction of the celebrated buybacks is offsetting dilution rather than shrinking the share count. The 10-year share-count change is only -1.25%, which for a company that has spent multiple billions on repurchases over the period is shockingly little compounding per share. The new CEO Ariane Gorin is competent but unproven at the consumer-marketing problem (she came from B2B); the prior CEO Peter Kern was a turnaround specialist, not a long-cycle operator. The IAC/Liberty/Diller governance overhang has eased but the dual-class structure remains and is a minor red flag. The acquisition history (trivago, HomeAway, Orbitz) is a graveyard of overpaid deals whose write-downs do not show up in current ROIC but did show up in the 10-year average ROIC of 1.31% — that is not a typo. One-point-three-one percent. This is not a high-quality compounder; it is a recovering one.

4. What bulls are extrapolating that won't hold. Bulls are extrapolating four things. First, that platform-consolidation margin gains continue indefinitely — they don't; one-time benefits are by definition one-time. Second, that AI is a productivity tailwind for Expedia (cheaper customer service, faster developer velocity); it is, but it is a much bigger tailwind for Google and the AI-assistant entrants who can disintermediate the booking surface entirely, and the net effect is negative. Third, that Vrbo will recapture share from Airbnb; six years post-acquisition, the trend is still wrong, and Airbnb has a stronger brand among the next generation of travelers. Fourth, that B2B can carry the whole valuation; B2B is structurally good but its 25% revenue contribution does not justify a 28x P/E on the consolidated business if the consumer side is shrinking.

5. Valuation trap (multiple compression / regime change). EXPE trades at 28.1x trailing P/E and 15.1x EV/FCF. A reverse-DCF embeds 4.7% perpetual growth — modest, but only if you believe the consumer business is stable. If you believe Google AI Overviews compresses consumer revenue 1-3% per year for several years while marketing intensity rises, the right multiple on the consolidated business is closer to 10-12x EV/FCF, not 15x. That alone implies a 30-40% derating from current price even before any earnings cuts. Layer on a 10-15% earnings reset from marketing-efficiency loss and the math becomes brutal. The 10-year average P/E of 311x in the scorecard is meaningless (post-COVID earnings collapse skews it), but the structural setup mirrors what happened to Yelp, TripAdvisor, and Groupon when Google started serving their content directly: those stocks compounded negatively for a decade.

If I am right, the stock could be worth $130-150 within 3 years.

Lollapalooza Bias Check

Several Munger biases are pulling on me as I write this analysis, and I want to name them explicitly.

Anchoring. I am anchored to the scorer's base IV of $273.85 and to the headline that EXPE 'looks reasonable' near current levels. The IV is a model output, not a fact, and the scorer flagged maintenance-capex uncertainty wide enough to widen the IV range materially. If maintenance capex is higher than assumed (entirely plausible given AI re-platforming and customer-acquisition reinvestment needs), the true base IV is lower than $273. I should weight the low IV ($157.50) more heavily than the optical midpoint suggests.

Recency. Q3 2025 was a strong quarter for travel broadly and EXPE specifically; the post-COVID recovery has been better than feared. Recent strength makes the AI-disintermediation thesis feel theoretical. But the analogous setup — Google starting to serve content from a category directly — preceded multi-year stock declines for Yelp (2014 onward), TripAdvisor (2014 onward), and Groupon (more or less continuously). The base rate for 'platform partner becomes platform competitor' is not friendly, and recency bias makes me discount it.

Authority / social proof. Expedia is widely owned by quality-oriented institutional investors and frequently appears on 'underrated travel platform' screens. The fact that smart people own it makes me want to find a reason to like it. But quality investors have also been wrong about TripAdvisor and Yelp for a decade in similar setups; institutional ownership is not a thesis.

Confirmation. Once I noted that 10-year ROIC is 1.31%, I started looking for confirmation that this is a structurally subpar business. I should challenge that: forward ROIIC is the better metric, and on the new platform, incremental returns may be substantially higher than the lookback. The scorer notes 'Net capital return period; ROIIC not meaningful,' which is honest — we genuinely do not know forward ROIIC, and the bear case overweights the historical figure.

Commitment / consistency. I have already characterized the moat as 'narrow,' and that initial framing is now coloring downstream sections (industry, inversion). I should test whether the B2B segment alone deserves a 'narrow' moat verdict (it might deserve 'narrow-to-wide').

Incentive-caused bias. The output format rewards a clear recommendation. 'Hold' or 'too uncertain to call' is a less satisfying answer than 'Avoid' or 'Buy,' and I have to resist the pull to write a more decisive headline than the evidence supports. The honest answer here is Hold with low conviction, and the headline reflects that.

Net effect of these biases: my analysis is probably slightly too bearish on the moat and slightly too bullish on the IV. Both adjustments push toward the same recommendation — Hold — at a price below current.

10-Year Outlook

Same fundamental business model in 10 years? Probably not, in an important sense. The 'OTA' label may persist, but the unit of work shifts from 'website where I shop hotels' to 'API that an AI assistant calls.' Expedia has been preparing for this with its Rapid API, white-label B2B, and unified platform stack — the bet is that even in an AI-mediated future, somebody has to actually hold the merchant-of-record relationship, the supply contracts, the loyalty ledger, and the customer-service operation, and Expedia is well-positioned to be that infrastructure layer. That is plausible. It is also unprovable today.

Customer base larger? Yes in the sense that travel TAM grows; uncertain in the sense that Expedia's share of bookings could shrink as Google, Airbnb, and AI-assistant entrants take direct distribution. B2B customer base almost certainly larger; B2C uncertain.

Profit per customer higher? Modestly higher if One Key drives repeat booking rates and if the loyalty cross-redemption raises the per-customer LTV the way Marriott Bonvoy does for Marriott. Lower if customer-acquisition cost rises faster than per-booking margin, which is the AI-Overview risk.

Moat wider? Hard to argue yes. The platform consolidation widened the operational moat, but the surrounding competitive environment (Google's role, Airbnb's expansion, agentic AI) is widening the threats faster than the moat. Net likely flat to narrower.

Single biggest threat? Google plus AI assistants disintermediating the consumer search funnel, compressing marketing ROAS structurally over a 3-7 year window.

The combination — uncertain business model evolution, uncertain customer growth on the consumer side, uncertain incremental ROIC, and a known structural threat — argues that I cannot draw a high-confidence picture of EXPE in 2036. I can imagine a world where B2B carries the company to a higher valuation, and I can imagine a world where consumer melts and the multiple compresses to single digits. Both are credible.

CONFIDENCE: low

Position Guidance

• Recommendation: Hold
• Conviction: Low
• Target buy price: $175 (close to scorer's low IV of $157.50, where margin of safety becomes meaningful given the AI-search overhang and 1.31% historical ROIC)
• Target trim price: $400 (above the high IV of $413.16; if multiple expansion gets you here, the bull case is more than priced in)
• Position sizing: If owned, no larger than a 1-2% position given low conviction and structural-threat asymmetry. Not a starter position at $251.84 — wait for the high-$170s or a strong B2B-segment disclosure that materially de-risks the consumer-disintermediation thesis.
• Conditions to revisit: (1) Google publishes AI-Overview impact data and OTA click-through stabilizes; (2) EXPE separately discloses B2B unit economics that justify a re-rating of just the B2B stub; (3) sustained gross-bookings growth above industry rate for 4+ consecutive quarters; (4) stock trades below $180.