teardown 01airbed & breakfast, 2008
Airbnb’s 2008 pitch deck: the numbers chain everyone leaves out
Airbnb’s 2008 deck is copied for its slide count. The count is not what worked. Five figures across two slides form one continuous chain, each the input to the next: 2 billion trips, 560 million of them budget and online, a 15% share giving 84 million, a $25 average fee, $2.1 billion of revenue. It multiplies out. One footnote on it does not.
What the deck is
AirBed & Breakfast, San Francisco, 2008. Brian Chesky, Joe Gebbia and Nathan Blecharczyk had already run the product at South by Southwest in 2007 and at the Democratic National Convention in 2008 before this deck existed. It asks for USD 500,000. The round that closed in April 2009 was USD 600,000, from Sequoia Capital and Y Ventures.
Eleven slides in the version that circulates: welcome, problem, solution, market validation, market size, product, business model, adoption strategy, competition, competitive advantages, team. The two that carry the argument are five and seven, and they are the two this teardown rebuilds.
Three things it does well
Five figures, and each one is the input to the next
The market slide runs three circles left to right: 2 billion+ trips booked worldwide, sourced to the Travel Industry Association of America and the World Tourism Organization; 560 million of those budget and online, sourced to comScore; and 84 million as a 15% share of that. Fifteen percent of 560 million is 84 million, so the third circle is the second one multiplied.
The business model slide picks up the same 84 million and carries it on: $25 average fee, from $80 a night across three nights at a 10% commission, to $2.1 billion of revenue projected by 2011. Eighty-four million times twenty-five is two point one billion. It closes.
That is rarer than it sounds. Most decks state a market size, then state a revenue projection, with nothing joining them. Here a reader can walk the whole business in about ten seconds, and it survives the walk.
Nearly everything on it had already happened
The product had run at two real events with real bookings before the deck was written. The validation slide points at Craigslist listings and Couchsurfing membership, which are other people’s numbers rather than the founders’ forecast. A deck built from things that already happened is a much harder document to argue with than one built from intentions.
The whole business, in seven words
The cover carries the line “Book rooms with locals, rather than hotels.” No jargon, no category invention, nothing after it required to explain what the company is. It is the hardest slide in any deck and it is usually the one written last, in a hurry.
Three things that would fail today
The slide’s own small print does not match the number above it
Under the average fee, the deck sets out its working: $80 a night, at three nights. Under that, in the smallest type on the slide, it names its source: $70 is the average room price on AirBed & Breakfast. The model runs on eighty. Their own measured figure is seventy, and the slide does not say why the two differ.
Follow it through. At $70 across three nights a 10% commission is $21, not the $25 on the slide, and 84 million trips at $21 is about $1.76 billion rather than $2.1 billion. The gap is roughly a fifth of the headline number, and every input needed to find it is printed on the slide.
There may be a good answer: an average room price and an average booking value are different things once a guest books more than one room. The problem is that the deck does not give one. In 2008 nobody was going to run that division in the room. In 2026 the associate reading it has a calculator open, and the one slide built to prove rigour is the slide that breaks.
It states a position where it needs evidence
The claim is that this is the cheapest online option. In 2008 an investor took that on trust. Today the same slide needs the three prices that prove it, with dates, because the reader will otherwise open a tab and check.
Product screenshots doing a designed slide’s job
There is no typographic hierarchy: headline, body and caption are close enough in weight that the eye has nowhere to go first. In 2008 that cost nothing, because almost nobody’s deck looked better. The median seed deck in 2026 is templated and competent, so a deck that looks like this one reads as unserious before a word of it is read.
Three slides rebuilt
None of the original artwork appears here. The schematic view is our own diagram of what each slide carried; the rebuild is our work. Toggle between them.
Published decks only. We reproduce none of the original artwork: the “before” panels are reconstructions we draw ourselves.
01 · The cover
Their words, our setting. The line needed nothing except to be set at the size of its own importance.
02 · Market size, slide 5
Market and model on one slide, with the sensitivity stated. It shows you did the division before they did.
03 · Business model, slide 7
A slide almost nobody builds. It is the one that survives due diligence, because it answers the question before it is asked.
The law it proves
A number on a slide must be the input to another number on a later slide, or it comes off.enzwa · craft law 01
A deck full of unconnected figures reads as research. A deck where the figures multiply together reads as a business. The test is mechanical: take any number on any slide and find the number it feeds. If there is nothing downstream of it, it is decoration, and decoration in a financial section costs you credibility rather than buying it.
The companion finding is in the note on where a document gets read, which pairs this deck with UberCab’s from the same year. The two broke opposite rules and both raised.
Sources
Every teardown uses a publicly published deck. Schematics and rebuilds are our own work.