A companion piece to Episode 7 of the BMD marketing case-study series. Episode 6 closed the failures. This one opens the other half, what to do. The video tells the story in minutes. This is the full autopsy: what Airbnb actually cut, what the famous 95 percent number does not say, and the order to follow before you touch your own marketing budget.
Contents
The cut everyone quotes
In 2020 Airbnb cut its brand and performance marketing from 1.14 billion dollars to 482 million. That is 662 million dollars, 58 percent, gone in a single year. In February 2021, on the company's first results call as a public company, Brian Chesky said it out loud: what the pandemic showed is that they could take marketing down to zero and still have 95 percent of the same traffic as the year before, and they were not going to forget that lesson.
That sentence has justified more budget cuts than any other sentence in modern marketing. It turns up in decks in Riyadh and Dubai with the company logo on the slide and the second half of the story missing.
Here is the part that never makes the slide. Airbnb's own filings show sales and marketing spending of about 1.19 billion dollars in 2021, 1.52 billion in 2022, about 1.8 billion in 2023, 2.1 billion in 2024, and about 2.6 billion in 2025. In the first half of 2026 that line grew about 32 percent against 17 percent revenue growth, and the filings put the increase down to paid growth initiatives. Airbnb spends more on marketing today than it ever did before the famous cut.
Both facts are true. The lesson people took from the first one is the reason the second one surprises them.
The machine before the cut
Airbnb in 2019 was eleven years old and had already become a verb. Revenue of 4.8 billion dollars, a listing base no competitor could copy in an auction, and a category the company had defined well enough that people searched for it by name. In its IPO filing Airbnb said roughly 91 percent of its guests arrived through direct or unpaid channels.
That last number is the whole company in one line. A two-sided marketplace compounds: every host is a reason for a guest to arrive, and every guest is a reason for another host to list. Ten years of that builds something a rival cannot outbid, because it was never for sale in the first place.
Then there is the fact that gets left out of the story entirely. 2019 was the year Airbnb spent the most it had ever spent. Brand and performance marketing rose 71 percent in one year, from 666 million dollars to 1.14 billion. Hold that number next to the cut. A company that has just raised its ad budget by 71 percent has far more room to cut than a company that has not.
The decision was not a strategy
In the first weeks of March 2020 the travel category stopped. Airbnb lost most of its business in about eight weeks. Revenue for the full year fell 30 percent, from 4.8 billion dollars to 3.4 billion. In May the company let 1,900 people go, about 25 percent of its workforce. It raised 2 billion dollars in debt at rates it would not have accepted a year earlier, and it had an IPO to reach in December.
Marketing was switched off inside that. Nobody sat in a room, weighed brand against performance on the evidence, and chose. The company stopped bidding for travel demand because there was no travel demand to bid for.
That is worth saying plainly, because every retelling since has made it sound like an experiment somebody designed.
What came back
Traffic returned to about 95 percent of the previous year's level with almost nothing behind it. Airbnb read that as a finding rather than a fluke, and it changed where the money went when the money came back.
On 22 February 2021 the company launched Made Possible by Hosts, its first global brand campaign in five years. The shape of the budget changed with it: more public relations and brand at the top of the funnel, less bidding at the bottom. Performance marketing, the part that had fallen 541 million dollars, did not come back to its 2019 share.
The results after that are not in dispute. 2022 revenue was 8.4 billion dollars, up 40 percent. Net income was 1.9 billion dollars, the first profitable full year in the company's history. Free cash flow was 3.4 billion dollars.
So the strategy worked. The argument is about what it proves.
The scoreboard
| Metric | Figure |
|---|---|
| Before the cut | 2019 revenue 4.8 billion dollars; brand and performance marketing 1.14 billion, itself up 71% in a year from 666 million; about 91% of guests arriving direct or unpaid |
| The cut | Brand and performance marketing down 662 million dollars (58%) to 482 million in 2020; performance marketing alone down 541 million; brand marketing down 121 million |
| The line most people miss | Total sales and marketing fell 28%, from 1,621 million dollars to 1,175 million, because most of that line is payroll and field operations, not media |
| The context | 2020 revenue down 30% to 3.4 billion dollars; net loss 4.6 billion, including about 2.8 billion of IPO stock compensation; 1,900 jobs cut in May 2020 |
| The claim | Chesky, February 2021: marketing down to zero and still 95% of the prior year's traffic |
| The payoff | 2022 revenue 8.4 billion dollars, up 40%; net income 1.9 billion, the first profitable full year; free cash flow 3.4 billion |
| Where it went after | Sales and marketing about 1.19 billion dollars (2021), 1.52 billion (2022), about 1.8 billion (2023), 2.1 billion (2024), about 2.6 billion (2025); up about 32% in the first half of 2026 against 17% revenue growth |
Marketing as a share of revenue fell from 23.7% in 2019 to 14.2% in 2020, then climbed back to about 19% in 2024 and about 21% in 2025. The 95 percent is a traffic comparison reported by the company, not an audited figure, and it was measured in a year when the entire travel category had collapsed.
The autopsy: the number is real, the lesson is borrowed
"Airbnb cut its marketing in half and grew anyway." The number behind that sentence is accurate. Three things it does not say are the reason most companies that copy it get hurt.
The first is that 662 million dollars is the advertising budget, not the marketing budget. In the same filings, total sales and marketing fell 28 percent, from 1,621 million dollars to 1,175 million. The gap between the two figures is payroll and field operations. "Airbnb cut marketing by 58 percent" and "Airbnb cut marketing by 28 percent" are both correct, and they answer different questions. Quote the first number to your board and then apply it to the second line, and you will not cut media. You will cut people.
The second is that 2020 was not an experiment. An experiment needs a control group, and in 2020 there was no part of the world where travel carried on normally while Airbnb kept bidding. Auction prices fell at the same time, so the cost of the traffic Airbnb stopped buying fell too. The 95 percent is a real number about a year that cannot be run again.
The third is that the cut did not stay cut. Sales and marketing went from 1,175 million dollars in 2020 to about 1.19 billion in 2021, 1.52 billion in 2022, about 1.8 billion in 2023, 2.1 billion in 2024 and about 2.6 billion in 2025. Skift read the filings from 2020 to 2026 and found the shape of it: brand campaigns rebuilt in 2022 and 2023, budget rotated back into search marketing in 2024 and 2025, then paid growth initiatives in 2026. The company famous for cutting marketing has quietly rebuilt the engine.
There is one more thing worth knowing about the 90 percent direct-and-unpaid line, which Airbnb has repeated for years. It describes how a visit is counted, not what caused it. Somebody who watches a brand film in March and types the company name in July is recorded as direct. Brand spending manufactures direct traffic. A statistic that counts the effect as the cause will always flatter the strategy that produced it.
So what actually happened? Airbnb spent ten years building an asset: a name people search for by name, a host base nobody else had, and a category it defined. In 2020 it stopped paying to reach people who were going to arrive anyway, and found out how many of them there were. The cut revealed the asset. It did not build one.
Quibi had 1.75 billion dollars and no asset underneath it. New Coke had an asset so deep it survived the company attacking its own product. Peloton had eighteen months of scale and read it as depth. Airbnb is the first case in this series where the answer to the question is yes, and that is exactly why it opens the half about what to do.
Bought demand and owned demand
Define the two before you argue about the budget.
Bought demand stops the day the invoice stops. Owned demand arrives because of something you built: a name people type, a customer who comes back, a host who tells another host, a piece of a category you defined first. Almost every company has both. Almost none can tell you the split.
The number worth knowing is not your return on ad spend. It is what share of the people your ads reach were coming anyway. No platform will hand you that figure, because the last click belongs to the platform either way, and a click you would have got for free looks identical in the dashboard to a click you genuinely bought.
That is why this is a department question and not a channel question. Somebody has to own the measurement, run it on a schedule, and be allowed to report a number that makes last year's budget look wrong. An agency will not do that to its own retainer. A dashboard cannot do it at all. It is the kind of thing a marketing function is for, and it is the first thing missing in most of the 40+ departments we have worked inside.
The right order: measure, then cut
Test incrementality on a slice, never on the whole budget. Split your regions into two groups, hold the spend back in one of them for two to four weeks, and measure total orders rather than platform-attributed ones. It costs a fraction of the budget and it answers the only question that matters: what happens to revenue when the ads stop.
Name the asset that will catch the demand before you cut into it. Write it down before the test runs: branded search volume, direct share of sessions, repeat purchase rate, referral rate. If the honest answer inside the room is that the ads are what brings people, then there is no asset to cut into. You are renting demand, and that is worth saying out loud before you stop paying the rent.
Spend the saving on the thing that compounds, then run the test again next year. Airbnb put its saving into brand and public relations, and its own answer still expired. The company that said it could take marketing to zero now spends about 2.6 billion dollars a year on it. The finding was true for its market at that moment. Markets move.
The same three steps shrink to a Riyadh-sized budget without changing shape. A store turns off Google Ads for one week in a slow month, sees flat revenue, and calls it proof. A clinic's entire lead flow sits inside one agency's ad account, and nobody has ever checked how often the clinic's own name gets searched. A restaurant group's brand, in practice, is a delivery app's ranking. Switching off the spend in those three cases does not reveal an asset. It reveals that there was never one there.
The takeaways
A budget cut is not a strategy. Airbnb's cut worked because of what was already underneath it, and the ten years that built it are the part nobody copies.
Know which line you are quoting. Brand and performance marketing fell 58 percent. Total sales and marketing fell 28 percent. Apply the first number to the second line and you cut people, not media.
2020 was not a test. No control group, collapsed demand, collapsed auction prices. Treat the 95 percent as a finding about one extraordinary year, not as a law.
A cut reveals an asset, it never builds one. Before you cut, name what will carry the demand. If the answer is the ads, you are renting.
Any answer about spend expires. Airbnb re-measured and rebuilt. The question is not what Airbnb decided in 2020, it is when you last ran the test yourself.
Frequently asked questions
Did Airbnb really cut its marketing budget by 662 million dollars?
Yes, on one specific line. Brand and performance marketing fell from 1.14 billion dollars in 2019 to 482 million in 2020, a drop of 662 million or 58 percent, with performance marketing accounting for 541 million of it and brand marketing for 121 million. Total sales and marketing, which also carries payroll and field operations, fell 28 percent, from 1,621 million dollars to 1,175 million. Both figures come from the same filings.
Did Airbnb prove that performance marketing does not work?
No. It found that a large share of the traffic it was paying for in 2019 was arriving anyway, in a company with about 91 percent direct and unpaid traffic, after a year in which it had raised its ad spend by 71 percent. That is a finding about Airbnb's own spend at its own level of brand strength, in a year when travel had stopped. It is a reason to measure your own incrementality, not a reason to assume the answer.
Is Airbnb's 90 percent direct traffic figure real?
It is a real measurement of how visits are counted, and it does not say what caused them. A person who sees a brand campaign in March and types the company name in July is recorded as direct. Brand spending produces direct traffic, so the figure will always look good for a brand-led strategy. Read it as a description of the funnel, not as proof of what the budget did.
Does Airbnb still spend less on marketing than it used to?
No. Sales and marketing went from 1,175 million dollars in 2020 to about 1.19 billion in 2021, 1.52 billion in 2022, about 1.8 billion in 2023, 2.1 billion in 2024 and about 2.6 billion in 2025, and grew about 32 percent in the first half of 2026 against 17 percent revenue growth. As a share of revenue it fell to 14.2 percent in 2020 and has climbed back to roughly 21 percent.
How do I find out how much of my own paid traffic is incremental?
Run a geo holdout. Split your regions into two comparable groups, pause the spend in one group for two to four weeks, and compare total revenue between the groups rather than the numbers inside the ad platform. Decide in advance what result would change the budget. The test costs a fraction of the spend it is testing, and it is the only way to see demand you are buying twice.
Should a mid-market company copy what Airbnb did?
Copy the measurement, not the cut. Airbnb could afford the answer it got because it had a decade of owned demand underneath the spend. The useful move for a company without that decade is to find out what it actually has, on a slice of the budget, before it decides what to stop paying for.
This article is part of BMD's marketing case-study series. Episode 7 opens the half about what to do. Episode 8 is the campaign that sold a men's product by talking to the women who buy most of it, and lifted unit sales 125 percent in a year. That is Old Spice.
Sources and further reading
Airbnb, Inc. annual reports on Form 10-K, fiscal years 2020 through 2025, and the 2020 registration statement: revenue of 4.8 billion dollars (2019) and 3.4 billion (2020); total sales and marketing of 1,621 million dollars (2019) and 1,175 million (2020), then about 1.19 billion (2021), 1.52 billion (2022), about 1.8 billion (2023), 2.1 billion (2024) and about 2.6 billion (2025); the brand and performance marketing line of 1.14 billion dollars (2019) and 482 million (2020), including the 541 million dollar reduction in performance marketing and the 121 million dollar reduction in brand marketing; the 71 percent increase in marketing spend during 2019 from 666 million dollars; and the disclosure that approximately 91 percent of guests arrived through direct or unpaid channels. WARC and Campaign, on the 541 million dollar performance marketing cut and the permanence of the shift to brand. Marketing Week, interviews with Airbnb's chief financial officer on the move from performance to brand building and on the later disciplined return of spend. Airbnb newsroom: Made Possible by Hosts, the first global brand campaign in five years, launched 22 February 2021, and the fourth-quarter and full-year 2022 results, 8.4 billion dollars of revenue up 40 percent, net income of 1.9 billion dollars in the first profitable full year, and free cash flow of 3.4 billion dollars. Brian Chesky's remarks on taking marketing to zero and retaining 95 percent of the previous year's traffic, made at the company's annual results presentation in February 2021 and widely reported at the time. Skift, August 2026, a reading of Airbnb's filings from 2020 to 2026 showing brand campaigns rebuilt in 2022 and 2023, budget rotated into search marketing in 2024 and 2025, and paid growth initiatives cited in 2026.
About BMD
Most companies don't have a marketing problem. They have a marketing department that was never built. BMD is a boutique consultancy that installs structured, measurable marketing departments inside mid-market companies across the GCC. We don't run your campaigns, and we don't hand you a strategy deck and leave. We build the operating system: the structure, the measurement, and the ownership that turn marketing into a function leadership can rely on. The method is the BUILD framework, published and practiced: a book, an online program, a community of Gulf founders and marketers applying it, and diagnostics that replace assumptions with measurement. Delivered in Arabic and English, founder-led.
Redha Alayesh
A marketer with a software engineer's discipline and a scientist's mindset. Across 40+ marketing departments in the GCC, he built the BUILD framework to solve the problem he kept finding: capable marketers trapped inside companies that never built them a department.