Don’t Start a Travel Startup

Don’t Start a Travel Startup. Start a startup inside a travel power struggle. Fourteen layers of advice from fourteen years of watching where the bodies are buried, from the founder of Skift. The best advice I can give a travel startup founder, fourteen years into building Skift, is this: don’t start a travel startup. Start a startup inside a travel power struggle. Travel is a $10 trillion industry, and every pitch deck leads with that number as though size equals opportunity. It does not. That spending is spread across a fragmented industry shaped by local rules, difficult operations, different payment systems, and emotional customer decisions. Founders see the size. They underestimate how difficult it is to reach. You don’t get to skip the complexity and keep the size. I have been watching travel startups launch, pitch, raise, pivot, and die for almost a decade and a half, from the front row. Over those years, Skift developed several ways to understand the industry. I have adapted fourteen of them for founders and organized them into three parts. Most come back to one question: where does the next dollar go in travel, and who decides? Before you build anything, learn who holds what. Permission to Exist. The question before all questions: what structurally gives you the right to be here? Skift’s Permission to Exist framework, which I wrote about a year ago, judges whether a company deserves its place in this industry on five questions. Can the places where travel gets found and booked actually see and sell what you have? When the trip goes wrong, are you the one the traveler blames? Blame sounds like a liability. It is proof you matter. Does anyone ask for you by name before the channel picks for them? Do you own the booking itself — the checkout, the payment, the customer record? And does the industry’s plumbing run through you? Every major change in distribution forces companies to answer these questions again. The internet did it to travel agents. The OTAs did it to hotels. AI agents are doing it now to everyone. The questions remain the same. What counts as a good answer changes. A startup cannot prove these answers yet, but it can decide which ones it intends to own. Use the five questions as a design brief. Choose which ones you will win, and make every trade-off deliberately. Airbnb has so far kept its inventory out of the AI agents — while Booking and Expedia built plug-ins inside ChatGPT, Airbnb said no and went off to build its own — and it can afford that refusal because travelers come to Airbnb by name. A small reseller locked out of those same agents is in the exact same spot — except Airbnb chose it, and the reseller didn’t. That’s the whole difference. Then compare those choices with the scale of the incumbents you are taking on. Booking Holdings spends $8 billion a year on marketing, more than all the venture money that goes into travel startups globally in most years. “We have better UX” won’t help you with any of the five questions — experience can make a product loved, but on its own it rarely gives you structural control. If your company has no real answer to at least one of the five, you don’t have a company. You have a feature. Features get copied, or absorbed, which is what happened to Hipmunk — anyone remember Hipmunk? — the best-loved interface in travel search, bought by SAP Concur and switched off just over three years later. Better UX was not permission. 1. Discoverability — can agents see and read your inventory? 2. Accountability — who gets blamed when the trip goes wrong? 3. Brand preference — does anyone ask for you by name? 4. Transaction ownership — who owns the booking itself? 5. Structural dependency — does the plumbing run through you? A WEAK ANSWER YOU CHOSE IS A MOAT. A WEAK ANSWER FORCED ON YOU IS A WARNING. Pick the Layer, Not the Logo. Travel is not one market. It is a layered system, and each layer has different economics, incumbents, and margins. Most founders say “we’re in travel” when they mean they built a prettier interface on top of someone else’s economics. Know which floor you are attacking, and know whether that floor actually captures margin. Inspiration is free, and the money is very likely three floors down. Consumer inspiration Where people dream. Inspiration is free, the money is three floors down. Search & planning Where intent forms. Google owned this floor for twenty years and charged rent on it. The AI answer engines are now fighting over the lease. Booking & payments Where money moves. The knife fight. Distribution rails Reservation networks, APIs, wholesalers, channel managers. The toll roads. Operations Hotels, airlines, airports, staffing. Where the promise gets delivered, or doesn’t. Capital allocation Owners, lenders, sovereigns, REITs. They decide where the money goes. Government & destination systems Visas, taxes, tourism boards. The scaffolding under everything. The Toll. A $500 hotel night leaves the hotel with anywhere from $350 to $475, depending on which rail the booking traveled through. Someone collects the difference, and they have been collecting it for decades. The entire history of travel technology is a fight over who collects the toll — the OTA commission, the fees charged by the big airline reservation networks, the card interchange, the app store cut, the metasearch click. Before you build anything, map the rail. Becoming a new toll collector means fighting incumbents with decades of supply relationships and fraud workflows baked in. Helping suppliers avoid the toll is more interesting, though the OTAs are quietly becoming everyone’s backend anyway. The better position: be the rail itself. The question for the next decade is which of these tolls survive when AI handles the booking. The hatched zone is the toll. The forward question: which toll booths survive when AI does the booking? Read Their Capital Allocation. What incumbents do with their money, as opposed to what they say on the record, tells you where the gaps are. Marriott spends 2.3% of revenue on technology, Hilton 2%, and both claim to be technology platforms on earnings calls. Booking spends 3.4% on disclosed IT against 30% on marketing, a 9-to-1 ratio, while telling investors it is investing heavily in AI. The biggest travel companies return 60 to 75% of free cash flow through buybacks. They are betting, with their actions, that the current model is permanent. If you believe otherwise, build where their capital is not going. One warning before you do: an empty space is not automatically an opening. Sometimes capital stays away because the economics are bad, the market is too small, or nobody will pay. Companies that claim to be technology platforms on earnings calls, allocating capital like utilities. The gap might be your opening — the next plate is the test. Build Where Incumbents Are Structurally Conflicted. The best openings are where the big players cannot act because their own economics prevent them. For every gap in their spending, ask why it exists. Can the incumbent enter that area whenever it chooses, or would doing so damage its existing business? Find the contradiction, build inside it. Booking cannot help hotels leave Booking. To beat you at this game, the incumbent would have to break its own business first. That is a position the incumbent cannot easily copy. OTAs Airlines Hotel brands DMOs To compete with you here, they would have to break themselves first. Where to enter, and whom to serve. The Consumer Travel Startup Is Dead as a Venture Thesis. Consumer travel businesses can still be built. What no longer works is raising venture money to buy customers at scale. Raise capital, spend it on acquisition, achieve scale, exit — that playbook fails in consumer travel. Google controls much of travel discovery. Booking and Expedia control much of the booking process. Suppliers continue to push direct sales, and AI agents are beginning to answer travelers’ questions before they visit any of them. Every consumer travel startup of the last decade pivoted to B2B, got acqui-hired, or shut down. The companies that survived are infrastructure, the ones the traveler has most likely never heard of. The graveyard of travel startups is full of itinerary builders. Business Pain Over Traveler Pain. Travelers complain constantly. That does not mean they will pay you to fix it. Businesses will. The better wedge is almost always on the business side: hotels leaking margin through bad distribution, airlines failing at retailing after a decade of talking about it, destinations counting arrivals instead of profit. And in travel, the user, the payer, the supplier, and the beneficiary are often four different people. Your buyer is never “the travel industry.” It is one executive who controls a budget — a CMO buying marketing tech, an airline CFO funding AI, a head of revenue management shopping for software. The opportunity only matters if someone has a budget to pay for it. Find the executive who owns that budget and sell to them. The person who uses it, the person who pays for it, and the person you sell to are rarely the same person. Know Which Friction You’re Solving. The industry uses “friction” as a uniformly negative word. Remove friction, increase conversion — the premise of every Connected Trip pitch and every AI agent demo I have sat through. The Friction Framework, which I came up with, asks two questions: who is affected by the friction, and does it destroy value or protect it? That creates four types. Barrier friction stops people from buying, so remove it. Legacy friction wastes time and labor, so automate it. Discovery friction is part of the pleasure of planning, so preserve it. Competitive friction is the complexity your company has learned to handle better than others, so keep it. That is your moat. BarrierREMOVE IT Visas, missing routes, broken payments. China dropped visa requirements for most of Europe and inbound trips surged, no campaign required. The barrier fell and the demand walked in. LegacyAUTOMATE IT Fragmented supply, manual processes, disconnected inventory. This is where most good B2B travel startups actually live. DiscoveryPRESERVE IT The pleasure of research, comparison, anticipation. Compress the planning phase too aggressively and you are solving a problem many travelers do not want solved. CompetitiveKEEP IT ON PURPOSE Operational complexity as moat. I recently wrote about Turkey’s travel tech sector, forged in currency volatility and geopolitical disruption — and better adapted for where travel is heading than platforms built for stable Western markets. Demand Moves in Corridors, Not Markets. Travel demand does not spread evenly across the globe like water finding its level. It moves in corridors — durable flows shaped by flights, visas, diaspora, trade ties, payments, and events. When Ryanair started flying to Tirana in late 2023, fares fell and traffic surged. Albania did not suddenly become more appealing. It became easier and cheaper to reach, especially for Italians, the Albanian diaspora, and budget travelers from northern Europe. Founders should study these flows, not just countries. Corridors show where demand comes from. Markets are where the company operates. Build around the corridor, then expand market by market. Albania was a corridor being opened. The demand was already there, waiting for the flights. Win a Specific World First. Travel may look global, but companies still have to operate country by country, under different regulations, payment systems, and supplier relationships. Start with a clearly defined customer and operating environment. That might be boutique hotels in Morocco, Muslim family travel, or forecasting for city hotels around major events. The world can be a market or a travel corridor; what matters is that it has a clear customer and a distinct set of operating conditions. Become useful enough to matter there first. Then decide whether the next market is similar enough to expand into or different enough to require a new approach. A corridor is a specific world. Become indispensable inside the smallest rectangle first. Look Beyond the Usual Hubs. Travel tech is not concentrated only in San Francisco, London, Amsterdam, and Singapore. Some of the strongest companies are being built elsewhere, often with less attention and less capital. Turkey is home to Hitit, one of the world’s largest airline reservation technology companies. Montreal produced Hopper, Plusgrade, and several other travel infrastructure businesses. Indian travel tech companies are increasingly acquiring Western firms. Founders should look beyond the usual hubs. The best opportunities may be in places where expertise is growing faster than investor attention. Fund it right, build something real, and watch where the customer is moving. Match the Capital to the Business. Venture capital works best when companies can scale quickly and a few winners can dominate. Travel is concentrated in booking, payments, and distribution, where the largest companies are already established. But the areas where startups usually enter — hotels, tours, and operations — remain fragmented by country, regulation, language, and supplier type. Those businesses are harder to scale, and there are fewer buyers when founders want to exit. The most durable travel tech companies were funded in different ways. Turkey’s airline Pegasus owns half of Hitit. In Canada, CDPQ helped Plusgrade expand while Softvoyage grew for 37 years without outside capital. In India, TBO and RateGain have used acquisitions to buy established Western companies — a pattern Skift calls Reverse Gravity. What these companies share is patience. They built infrastructure that took years of integrations and trust before it paid off. That rarely fits the timetable of venture capital. The lesson is simple: match the money to the business. The wrong investors can force a good company to grow too fast, sell too early, or make decisions that weaken it. Be a Builder, Not a Renamer. Skift’s Say-Do Gap compares what travel companies say about AI with what they are actually doing. Hiring is one of the clearest signals. After reviewing more than 170 AI job listings across more than 30 travel companies, three types emerged. Builders create new capabilities. Reorganizers use AI to improve existing workflows. Renamers add AI to a title or product without changing much underneath. Many travel startups fall into the last group. The stronger companies combine AI with supply access, industry knowledge, and operating experience. AI can make those advantages more useful, but it cannot replace them. The first test is whether the company has built something real. The second is whether it understands how AI is changing who owns the customer relationship. TAP: STRIP THE AI LANGUAGE FROM YOUR PITCH DECK Does the company still make sense? If yes, the AI is a tool and you may have a business. If no, you have a technology demo dressed up as a company — the say-do gap, in miniature. AI Changes Who Owns the Customer. If your business depends on travelers finding your website and booking there, you are building for the past. Skift argued in 2015 that booking would eventually happen on any digital surface. AI agents are making that real. SiteMinder has opened 53,000 hotels to ChatGPT and Claude. Long Lake took Amex GBT private in a $6.3 billion deal built around AI servicing. Capital One acquired Hopper’s technology to add travel booking to its financial platform. The question is no longer only who handles the booking. It is who owns the AI memory behind it. Banks already know what travelers spend, where they spend it, and how often. But no one yet holds a complete, portable AI memory of what a traveler liked, disliked, requested, or experienced across a trip. When the agent books for someone and learns what they loved and hated — who owns that memory? That AI memory could become a new source of power in travel. The opportunity is probably not another consumer brand. It is the infrastructure that stores that memory and makes it available to agents, suppliers, and banks. For a founder, that is the most valuable unclaimed job in travel. Booking on any digital surface — we called it in 2015. The AI memory behind the booking is the next fight. Fourteen years ago, I started working on the idea that became Skift with no travel industry experience or connections. I believed the business of travel was underserved by the media covering it. That was right, but it was not enough. Skift found an underserved layer — business information — served a specific customer, and expanded from there. Each step required choosing the right problem, understanding where companies were spending, and becoming useful in one part of the industry before moving into the next. Travel is the largest experiential economy in the world, but it has far less influence than its size suggests. That weakness creates openings. Do not build beside the power struggle. Build where an incumbent cannot follow without damaging its own business.