Airbnb Turned Off Half a Billion Dollars of Marketing. Almost Nobody Stopped Coming.

In the spring of 2020 the travel industry went dark, and Airbnb did something that should have been suicidal for a company that ran on search ads: it stopped buying them. It suspended substantially all performance marketing — the search and online bidding that fed the funnel — and waited to see who showed up anyway.² The number that came back is the whole story. Traffic returned to 95% of the prior year's levels before the company spent a dollar to bring it back, and more than 90% of fourth-quarter traffic arrived direct or unpaid.³ Airbnb had been paying, year after year, for visitors who would have come on their own.

The official story is that Airbnb spent heavily to build a beloved brand, and the pandemic forced a clever pivot toward it. The truer story runs the other way: the brand was already load-bearing, and the marketing budget was mostly camouflage hiding how little of it was needed. The cut didn't create the moat. It revealed one that had been there all along.

The number Airbnb buried in its IPO filing

When Airbnb filed to go public in November 2020, it disclosed a figure that quietly explained everything. In 2019 — a normal year, the company at full marketing throttle — only about 23% of its traffic came from paid performance channels. The other roughly three-quarters arrived direct or unpaid: people who typed the name into a browser, opened the app, or found it through a search they were going to make anyway.¹ That is not the profile of a company that buys its demand. It's the profile of a company whose customers already know where they're going. The performance spend wasn't building the audience; it was bidding against itself for an audience it mostly already had.

Here is the mechanism, worked down. Performance marketing is a tax you pay on intent that isn't yet attached to your name. You bid on a keyword like 'apartment rental Lisbon' because the searcher hasn't decided who to rent from. But once your brand becomes the category — once 'I'll just Airbnb it' is how people describe the act before they've opened any app — intent and brand fuse. The customer skips the auction entirely. They don't search for the category; they search for you. And when that happens at scale, every dollar of performance spend is increasingly buying clicks from people who were going to arrive direct. You are paying to intercept your own free traffic. Cut the spend, and most of that traffic simply walks in the front door instead.

Notice the precision of the claim. Not 'people like us.' A noun and a verb. The brand had crossed the line where it stopped being a choice inside a category and became the category itself — the way you Google a question or Xerox a page. That linguistic shift is the moat made audible. It is the moment marketing stops being an expense that scales with growth and starts behaving like an asset that's already paid for.

The honest catch: the auction got cheaper too

The fair objection is that 2020 was a rigged experiment. When Airbnb pulled its search budget, so did everyone else — the entire travel industry collapsed at once. Booking.com, the hotels, the OTAs all cut their paid-search bids as demand cratered, which means the keyword auctions Airbnb was no longer competing in were also quieter, cheaper, and less crowded. Some of that 95% 'resilience' was a market-structure effect: not pure brand pull, but the absence of rivals bidding against it. A clean test would have isolated brand strength while competitors kept spending. This one didn't. So the proof is real but partly confounded — the moat was demonstrated in a year when the moat had unusually little to defend against.

But the confound cuts less deeply than it first appears, for one reason: the 23% figure predates the pandemic. In 2019, with every competitor spending freely and auctions at full price, Airbnb still drew the overwhelming majority of its traffic from direct and unpaid channels.¹ The pandemic didn't manufacture the direct-demand habit; it stress-tested one that the IPO filing shows was already in place. And the durability is in the years since. By 2024, Airbnb had pushed sales and marketing back up to $2.1 billion, 19% of revenue — but committed to holding that ratio steady, with the majority going to brand rather than performance.⁷ That is the tell. A company buying its demand sees marketing rise faster than revenue as it scales. A company that owns its demand spends a constant slice and lets the brand carry the rest.

It was a verb before it could afford ads

The instinct to treat the brand as something bought late, with IPO money, misreads the company's own beginning. In October 2007 two design-school roommates rented air mattresses on their San Francisco floor at $80 a night to conference-goers who couldn't find a hotel room.⁵ A year later, broke and ignored by investors, they printed novelty cereal boxes — Obama O's and Cap'n McCain's — and sold roughly 500 of each at $40 a box around the 2008 Democratic convention, raising about $30,000 to keep the lights on. The Obama O's sold out; they ate the leftover Cap'n McCain's themselves.⁶ None of that was performance marketing. It was a brand learning to make itself memorable before it had a budget to buy attention. The verb was always the product. The ad spend, when it finally arrived, was just renting reach the brand would later prove it didn't need.

So the real anatomy of Airbnb's moat is not a clever campaign or a beloved logo. It is a single structural fact, hiding in plain sight in a federal filing: the company was paying for demand it already owned, and only discovered the size of the gift when a pandemic forced it to stop buying. The brand didn't become a verb because Airbnb spent the most. It spent the most for years without realizing it had already won — and the day it found out, it never went back. The deepest moat isn't the one you build. It's the one you find you've been over-defending all along.