An aggregator is a business model in which a company gathers the supply of many providers, such as hotels, sellers or drivers, in a single place and owns the point where customers begin their search. It earns a fee on each match, and its power comes from being the first place customers think to go.
SqueezedAI squeezes it: an assistant that searches, compares and books for the customer could become the new starting point, and push the aggregator a step back from the people it serves.
How it makes money
Start with where a trip begins. An aggregator does not own hotels; it owns the moment a traveler types a city and a date. Hotels pay to be in front of that moment, and the aggregator's margin is the price of the introduction.
The model has a long-known soft spot: much of its traffic is rented. When customers arrive through search ads, the aggregator pays for each visit, at a price set by everyone bidding for the same customer. The cheaper path is the customer who opens the app directly. AI raises a new question about both paths: what if the customer never arrives at all, because an assistant does the searching?
Rented traffic and owned traffic
Booking Holdings' 10-K describes both kinds. On the rented kind it is frank: "Performance marketing costs to grow traffic to our platforms are variable because they are dependent on others' marketing spend in the same channels." Against that, the share of room nights booked by consumers coming directly to its platforms "was a mid-fifties percentage in 2025." The direct share is the part of the business no middleman can tax.
The agent in the filings
Both of the large online travel companies now name AI agents as a threat to the model itself. Booking warns that "AI agents may further evolve into full-service booking platforms, increasing competitive pressure and disintermediating OTCs," the online travel companies, Booking among them. Expedia expects generative and agentic AI to intensify competition from rivals who "may deploy AI-driven travel search, planning, and booking capabilities more effectively or rapidly than we can," and calls a failure to navigate this "agentic" revolution a material risk.
Booking is not waiting to find out. It offers "consumer-facing Gen AI capabilities across our brands, including trip planners, AI assistants to answer consumer queries, and a price comparison tool." The aggregator is trying to become the agent before an agent becomes the aggregator.
How it fails
An aggregator fails when someone else owns the start. Search engines have long been that someone, which is why performance marketing is a cost the aggregators cannot set for themselves. An agent that books on the customer's behalf would go further. The aggregator would still supply the rooms, but it would be selling to software rather than to a traveler, and software compares prices without habit or loyalty.
Our read
The aggregator's moat was never the inventory. Hotels list everywhere. The moat was habit: being the app a traveler opens without thinking. AI agents go after habit directly, because the new habit is asking the assistant.
The defense the filings point to is the direct customer: the traveler who comes on their own, signs in, and keeps a history there. That share gets more valuable every year the agents get better. For an aggregator, the board question is how much of its demand it would keep if the assistants sent nobody. For a supplier, the agents are a chance to reach customers without paying the old toll, and a risk of paying a new toll to someone else.
Questions
- What is an aggregator business model?
- A company gathers the supply of many providers in a single place, such as hotels on a travel site, and owns the point where customers begin. It earns a fee on each match, and its power comes from being where customers start.
- How do AI agents threaten companies like Booking and Expedia?
- An AI agent that searches, compares and books on the customer's behalf could become the new starting point. Booking's 10-K warns that AI agents may evolve into full-service booking platforms, and Expedia calls the agentic shift a material risk.
- How can an aggregator defend itself against AI agents?
- By owning more of its demand directly. Booking says a mid-fifties percentage of its room nights in 2025 came from consumers who went straight to its platforms, and it is building its own AI trip planners and assistants.
Sources
- Booking Holdings Inc. (SEC EDGAR), Form 10-K for 2025 (2026-02-18)
- Expedia Group, Inc. (SEC EDGAR), Form 10-K for 2025 (2026-02-13)
Every figure and quotation on this page traces to these filings. “Our read” is Stratrix’s judgment, drawn from them. Analysis and data, not advice.
More from Stratrix
“How much of your revenue comes from somebody who is also trying to replace you?”
See the pack — $499 →The Business Model Wizard
Make the seven choices that define your model, test your numbers against real companies, and meet the rival AI would build against you.
Free · about twelve minutesBuild your Model Thesis →Test a decision, free
Set a live decision against the companies that made it before you.
AcquisitionsThe Distance TestDivestituresThe Divestiture TestDiversificationThe Diversification TestMake or buyThe Make-or-Buy TestAll Access
Every Casebook, three Strategy Packs a seat each year and the Analyst, for teams that face these decisions all year.
$1,999 a year · $499 a seatWhat is included →New analyses, by email
The defining moves in business, checked against the record, as they publish.