Pairs with the Flywheel Designer Canvas — a ready-to-use strategy tool. Included with a subscription, or $1.99.
In October 2007, two designers who couldn't make rent put air mattresses on their floor and rented them to strangers in town for a design conference.6 A year later, to stay solvent, they hand-made roughly 1,000 boxes of election-themed cereal and sold them at $40 a box.7 Fifteen years after the air mattresses, the company they built could tell investors that 36% of its new hosts in a single quarter were people who had first arrived as guests.3 That last number is the whole story - the loop closing on itself. The question is not whether the loop is real. It is how far it actually reaches.
The official story is that Airbnb is a flywheel: a frictionless, self-reinforcing machine that, once spinning, grinds every challenger into dust. Almost every word of that is half-true. The loop exists and it is powerful. But Airbnb never calls it a flywheel - in its own filings it says 'network effects,' and on the earnings call its CEO reached for a narrower phrase: a 'self-growing effect.'3 The gap between 'flywheel' and 'self-growing effect' is the gap between a global monopoly and a business that has to win the same fight in every city, forever.
“36% of new hosts in the quarter were prior guests - it gives the network a self-growing effect.”3
How the loop actually turns over
The mechanism is elegant, and it is worth saying out loud because it explains why the growth has been so durable. A guest books a stay, has a good experience, and discovers that the spare room or the second property they own could earn money too. So the demand side quietly manufactures its own supply: a third of new hosts arrive already trained as customers, already trusting the platform, already needing no convincing.3 More hosts means more listings; more listings means more choice; more choice pulls in more guests; and a slice of those guests become the next batch of hosts. The numbers show the wheel turning: by the end of 2023 Airbnb counted more than 5 million hosts and over 7.7 million active listings, up 18% in a year, with hosts collectively earning more than $57 billion.4 Listing growth, the company noted, accelerated every quarter after its 2020 IPO.5 This is not marketing. It is a genuine compounding loop.
The cost of acquiring supply falls toward zero because demand recruits it: roughly a third of new hosts were already guests.3 That is what powered listings to 7.7 million by the end of 2023, up 18% year over year, with hosts earning over $57 billion that year.4 But notice what the formula doesn't say: it never specifies where. The loop only closes between guests and hosts who can plausibly transact - and that is almost always the same city.
Why the wheel doesn't spin across borders
Here is the part the flywheel metaphor hides. A true flywheel stores energy and releases it everywhere; Airbnb's loop releases its energy only locally. A new apartment listed in Paris adds nothing for a traveler searching in Tokyo - they will never book it, will never even see it as a relevant result. MIT Sloan Management Review made exactly this argument back in 2017: Airbnb's network effects are geographically local, not global, which limits the moat relative to platforms whose effects are truly worldwide.8 The consequence is brutal in its simplicity. Airbnb does not have one flywheel; it has thousands of small ones, one per market, each of which can be slowed, regulated, or out-competed independently of the others. The platform is enormous in aggregate and surprisingly fragile in any single place. A city council that caps short-term rentals isn't denting a global moat - it is jamming one wheel, and the global machine keeps spinning while that local one seizes up.
| The flywheel story | How it actually works | |
|---|---|---|
| Where value accumulates | Globally, cumulatively | Locally, city by city |
| A new listing helps | Every guest, everywhere | Only guests searching that market |
| The moat | Winner-take-all | Re-earned in each market |
| Main vulnerability | Effectively none | Local regulation, host multi-homing |
Isn't the loop strong enough that none of this matters?
The fair objection is that a moat made of thousands of local loops is still a moat - and a deep one. To beat Airbnb in any given city, a challenger has to assemble both sides at once, locally, against an incumbent that already has the listings and the guests there. That is hard, and it is why no global rival has dislodged it. There is a second, stronger point in Airbnb's favor: resilience. When COVID erased travel, gross booking value for the first nine months of 2020 fell to $18 billion — down 39% year-over-year on a like-for-like basis — against a full-year 2019 figure of $38 billion.1 Yet the business rebounded within two months, pulled back not by a management masterstroke but by guests spontaneously shifting to domestic and longer stays.29 A platform that recovers because its users redirect its supply for it is genuinely robust. But notice the honest qualifier underneath: the recovery was demand-led, not a designed pivot, which is precisely the point. Airbnb's strength is that the local loops are self-healing - and its limit is that they are local. A serviceable market the S-1 pegged at $1.5 trillion2 is not won once. It is won one city at a time, and defended the same way.
Every platform with a network effect will tell you it has a flywheel. The question that separates a real moat from a marketing slide is: between whom, exactly, does the loop close - and where? If new supply helps every customer everywhere, you have a global flywheel and a winner-take-all market. If new supply only helps customers in the same place, time, or category, you have something far more common and far weaker: a bundle of local loops, each of which can be attacked, regulated, or replicated on its own. Airbnb's loop is real and it self-heals. But it accumulates locally, which is why a global giant can be perpetually exposed to a single city council. Before you believe a flywheel story, draw the loop and check the radius.
The cereal boxes worked. The air mattresses worked. The loop where guests quietly become hosts works best of all - it is one of the cleanest self-growing engines in modern business, and the listing counts prove it.4 But a flywheel is a promise about geography: that the energy you build in one place spins everywhere. Airbnb's doesn't. It spins exactly as far as a traveler is willing to book, and no further. The genius was never a single global wheel that no one could stop. It was learning to start thousands of small ones - and the cost of that genius is that they can be stopped, one city at a time.
Flywheel Designer Canvas
A one-page canvas for mapping a business's flywheel: the reinforcing loop, how it was started, the second-order loops it spins off, the moat it creates, and how it could spin backward. Use it to diagnose whether you have a real flywheel or a funnel drawn in a circle — and to design one of your own.
Included with any subscription, or unlock this tool for $1.99. Get it → · See plans →
Sources
Where this comes from — the filings, records, and reporting behind it.
- 1Airbnb's S-1 was publicly filed November 16, 2020; as a private company it had raised approximately $6.4 billion in funding; gross booking value fell from $38B in 2019 to $18B in 2020, a 39% year-over-year decline driven by COVID-19.
- 2Airbnb's S-1 states its serviceable addressable market is $1.5 trillion ($1.2T short-term stays + $239B experiences); the filing notes the business rebounded within two months of COVID disruption, driven by domestic travel.
- 3On the Q4 2022 earnings call, CEO Brian Chesky stated that 36% of new hosts in Q4 were prior guests, and described this supply-from-demand dynamic as giving the network 'a self-growing effect.'
- 4By end of Q4 2023, Airbnb's host community surpassed 5 million hosts, active listings exceeded 7.7 million (up 18% YoY), and hosts collectively earned more than $57 billion in 2023.
- 5In Q2 2023, active listings grew 19% year-over-year to over 7 million total — Airbnb's highest count at that time; listing growth accelerated every quarter since the 2020 IPO.
- 6The air-mattress proof-of-concept occurred in October 2007 during an Industrial Designers Society of America conference in San Francisco; Nathan Blecharczyk joined as CTO and third co-founder in February 2008; the website airbedandbreakfast.com launched August 11, 2008.Wikipedia, Airbnb ↗ · 2024
- 7The founders hand-made approximately 1,000 limited-edition election cereal boxes ('Obama O's' and 'Cap'n McCain's') sold at $40 each; proceeds (variously reported as $30,000–$40,000) kept the company solvent long enough to apply to Y Combinator, which admitted them into Winter 2009 and provided $20,000 for approximately 6% equity.
- 8MIT Sloan Management Review (2017) argues that Airbnb's network effects are NOT winner-take-all: the value added by a new listing is geographically local, not global, which limits the 'flywheel' moat relative to platforms with truly global direct network effects.
- 9Airbnb's S-1 filing states that among resilient areas during COVID were domestic travel, short-distance travel, travel outside top 20 cities, and long-term stays.