Airbnb 2009: Surviving Before They Could Scale
Situation
It is early 2009. Airbnb is 8 months old. It has been through two failed launches (at the 2007 Democratic National Convention and the 2008 SXSW festival), a Y Combinator application that Paul Graham almost rejected, and months of near-zero revenue.
The three co-founders, Brian Chesky, Joe Gebbia, and Nathan Blecharczyk, are $20, 000 in credit card debt from the previous year's DNC air mattress launch. They have been selling novelty themed breakfast cereals (Obama O's and Cap'n McCains) to pay rent. The cereal scheme generates ~$30, 000 in cash, enough to survive Y Combinator.
At YC, Paul Graham gives them advice that changes everything: "Go to your users."
The New York City listings on Airbnb are performing poorly. Revenue is flat. Chesky and Gebbia fly to New York, knock on the doors of Airbnb hosts, and find the problem immediately: the photos are terrible. Hosts are using grainy cell phone photos that make their apartments look uninviting. Guests who can't trust what a space looks like won't book.
The solution Chesky and Gebbia devise is not scalable: rent a $5, 000 camera and photograph every NYC listing themselves. Spend a week visiting host apartments, taking professional-quality photos, and replacing the listing images.
The results are immediate: NYC bookings double within a week.
This is the moment Airbnb figures out the marketplace trust problem.
The decision moment
It is Q2 2009. Airbnb is growing in New York but still tiny, fewer than 1, 000 total listings globally. Three paths are in front of the team:
- Scale the photography program. Hire professional photographers in every city where Airbnb operates. This is expensive and operationally complex, it requires a marketplace company to run a local operations team in every market. It is completely non-scalable in the conventional startup sense. But it works.
- Build self-serve photo tools. Give hosts better tools to take their own photos: guidelines, editing tools, quality scoring. This is scalable but unproven, if hosts don't use it or can't execute, you're back to bad photos.
- Ignore the supply-side quality problem and focus on demand. Get more guests, more bookings, more revenue, and bet that supply quality will improve naturally as hosts see what the best listings look like.
The decision is not just about photography, it is about the fundamental marketplace philosophy: how much does a two-sided marketplace need to curate and manage supply quality to make demand-side trust possible?
You are Brian Chesky.
Key financial datapoints (for reference)
| Metric | Value (2009) |
|---|---|
| Founder credit card debt | ~$20, 000 |
| Obama O's cereal revenue | ~$30, 000 |
| Airbnb total listings (global, early 2009) | <1, 000 |
| NYC listings before photo project | ~50-100 |
| Revenue before NYC photo project | Near-zero |
| Revenue after NYC photo project (week-over-week) | ~2x increase |
| Professional photography rental cost (YC era) | ~$5, 000 |
| Airbnb 2009 total revenue | ~$200K |
| Airbnb 2011 total revenue | ~$30M |
| Airbnb 2019 total revenue | $4.8B |
| Airbnb IPO (December 2020) | $47/share; ~$100B market cap day 1 |
Frameworks invoked
- Two-Sided Marketplace. Airbnb has a chicken-and-egg problem: guests won't book without good supply, hosts won't list without guest demand. The critical insight from the NYC photo project: supply quality (not just supply quantity) is what enables demand. Trust in the supply is the variable.
- Trust Architecture. Airbnb's core innovation is not the booking platform, it is the trust layer that allows strangers to sleep in each other's homes. Every product decision (photo quality, verified IDs, reviews, host response rates) is a trust architecture decision. Bad photos are a trust failure, not an aesthetic failure.
- Product-Market Fit. The NYC photo project is product-market fit discovery in its most tactile form: Chesky and Gebbia go to the market themselves, find the friction point, solve it in the most direct way possible, and measure the result. The scalability problem comes later. First you have to prove the model works.
- Founder-Market Fit. The co-founders' willingness to do non-scalable work themselves, photographing apartments, sleeping on air mattresses, selling cereal, is what makes early marketplace companies succeed. Founders who insist on scalable solutions from day one don't learn what actually matters.
Discussion questions
- Paul Graham's advice to "go to your users" unlocks Airbnb's growth. But the photography solution is explicitly non-scalable. At what point does a marketplace company need to stop doing non-scalable things, and how do you know when that point is?
- The trust problem in Airbnb is bilateral: guests must trust hosts, and hosts must trust guests. The photography solution addresses one side. What are the equivalent trust-building interventions on the guest side, and why are they harder to solve with a product feature?
- Airbnb's model requires hosts to let strangers into their homes. The hotel industry is regulated, standardized, and familiar. What was the minimum viable trust signal that convinced the first wave of hosts to list, and is it the same signal that convinces the millionth host?
- The DNC launch in 2007 worked (sort of): the problem was political conference demand and no supply. The 2008 SXSW launch failed: the problem was the product wasn't ready. What's the difference between a failed idea and a failed timing/execution, and how do you tell which problem you have?
- By 2019, Airbnb has 7M+ listings and $4.8B in revenue. The photography program evolved into a professional photographer marketplace with 50, 000+ photographers globally. Was the non-scalable seed of that program necessary, or could a scalable version have achieved the same outcome faster?
The real outcome (revealed at session end)
2009: The NYC photography program doubles NYC revenue immediately. The team rolls it out to other cities with hired local photographers. It is operationally painful and expensive, but it works.
2010: Airbnb raises a $7.2M Series A from Sequoia Capital. YC's investment in Airbnb becomes one of the highest-returning investments in Y Combinator history.
2011: Airbnb reaches 1 million nights booked, and $30M in revenue.
2012: Airbnb faces its first major trust crisis: a host's home is trashed by guests. Airbnb introduces the Host Guarantee (up to $1M coverage) and verified IDs. Both are trust architecture decisions.
2020: Airbnb files for IPO during COVID, a pandemic that eliminates travel. Despite a brutal 2020, Airbnb goes public at $47/share in December 2020, reaching a ~$100B market cap on day one.
The lesson: "Do things that don't scale", Paul Graham's most famous essay, is embodied by Airbnb's photograph project. The non-scalable thing is not inefficiency, it is the only way to learn what actually creates trust in a new marketplace category. You can scale what works. You can't scale something you don't yet understand.
Sources
- Brian Chesky, various interviews on Airbnb's early history.
- Paul Graham, "Do Things That Don't Scale" (PG essays, 2013).
- Leigh Gallagher, The Airbnb Story (2017).
- Airbnb S-1 (2020).
- HBS case: "Airbnb: Democratizing the Market for Space" (2014).