Doing things by hand to find out what the product should do
Airbnb's founders learned what their product needed by doing the work in person. Manual effort teaches, as long as someone is keeping score.
Do things that won’t scale; it will teach you.
In early 2009 the company then called AirBed & Breakfast was in Y Combinator’s winter batch and barely growing. Its founders, Brian Chesky, Joe Gebbia and Nathan Blecharczyk, had kept it alive the previous autumn by selling novelty breakfast cereal boxes themselves during the presidential election. Looking at their listings, they noticed that many of the New York apartments were shown in dark, blurry phone photos. So Chesky and Gebbia flew to New York, knocked on hosts’ doors and photographed the apartments with a borrowed camera. Bookings in the city went up.
The first half of the line was advice Y Combinator gave many of its companies: in the early days, do things by hand that would be absurd for a large company. The second half is Chesky’s own emphasis, and it is the part that explains why. Standing in a host’s living room taught the founders things no dashboard could. They learned how hosts priced their rooms, what worried them about strangers, why some listings never booked. Several of those lessons later became features, including a professional photography program.
Manual work teaches because it puts the founder in direct contact with the problem. A founder who personally onboards the first fifty customers hears every question, sees every confusion and learns which steps matter. Automating too early locks in guesses before they have been tested.
The advice has two failure modes. The first is when manual effort hides the absence of demand. A founder who personally drives every sale can produce growth that disappears as soon as he stops, and the numbers will look healthy until then. The work has to be read as an experiment: what happens with the help, and what would happen without it.
The second is when the manual phase never ends. Some companies drift into being service businesses with software attached, because the handwork keeps customers happy and nobody sets a date to replace it. The margins of that business are very different from the one the founders pitched. Doing things that do not scale is meant to be a phase with lessons and an exit.
The practical version is to keep notes. Each thing done by hand should produce a finding: what the customer needed, how long it took, what could be built to do it. A founder who flies across the country to take photographs is doing research, and research is only useful if someone writes down the results.