Build what you would use today
Being the first user removes the guesswork. It also narrows the view.
Just build something that you’d want to use today, not something you think people could use somehow.
Drew Houston began writing what became Dropbox on a bus out of Boston, annoyed at having left his USB drive at home. He needed the thing he was building, that week, and he could tell within minutes of using a prototype whether it worked. That is the situation Graham describes. The sentence itself circulates as his without a traced original, but its argument is the one he made in his April 2010 essay Organic Startup Ideas, which tells founders to ask what they wish someone would make for them.
The two halves of the line name two ways of arriving at an idea. “Something you’d want to use today” starts from a present, specific need that the founder feels personally. “Something you think people could use somehow” starts from an imagined user and a guess about what that user lacks. The word “somehow” carries the criticism: the founder cannot say exactly how, because the need is not theirs and has not been observed.
The practical advantage of the first route is feedback. A founder who is also a user knows at once when a feature is clumsy, and can fix it that afternoon. A founder building for imagined people has to wait for real ones, and by then months of work may rest on a wrong guess. Picture a team of students building a calendar for busy parents, none of whom has ever had to coordinate a school pickup. Every design choice is a guess, and the guesses tend to reflect what sounds useful in a meeting rather than what a tired parent reaches for at seven in the morning.
The method has a known blind spot. Founders’ own wants are not typical. A group of young programmers wants what other young programmers want, such as better developer tools. Those markets are real but crowded, and many large problems sit in fields the same founders never touch, hospital billing or freight scheduling for instance. Building only for oneself leaves those problems to people with less technical skill.
The fix is closeness rather than guessing. A founder who has worked inside an industry, or who spends enough time with its people to feel their frustrations directly, can build for them almost as well as for themselves. What the advice rules out is the distant version: a market chosen from a report, with users described in general terms and no one on the team able to say what the product must do by Monday.
“Today” is the part most easily skipped. An idea that the founder would use in some future life, after the right job or the right family, is still a guess. The test is present tense.