Learn startups from Paul Graham

Keep a problems log for two weeks: every time something is annoying, slow or missing in your work or life, write one line.

Graham says the way to get startup ideas is to look for problems, preferably your own, and that the right verb is "notice", not "think up".

Pick the three most promising problems and talk to ten people who have them. Ask how they deal with it now and what it costs them.

Michael Seibel lists "not knowing where your first users will come from" as a top first-time founder mistake. In the video he says it is odd to pick a problem when you don’t know anyone who has it, including yourself.

For each idea, write who wants it urgently. Prefer a small group who need it badly over a big group who would find it nice.

Graham says good ideas are usually narrow and deep, "like a well": intense demand from a small group first.

Ask of every idea: would I use this myself if I hadn’t built it?

When user growth is slow, Graham says the product is often the real problem, and asks founders: "Would you use this yourself, if you hadn’t built it?"

Read the 18 mistakes and cross out any idea that is a marginal niche or a copy of something that exists.

"Marginal Niche" and "Derivative Idea" are on Graham’s list of mistakes that kill startups, along with having no specific user in mind.

Write a one-page spec for a version you can build in weeks, then cut every feature your first ten users won’t miss.

Seibel’s MVP advice: time-box the spec, write it down, cut features hard, and aim for weeks, not months.

Set a launch date three to six weeks out and launch on it, even if it embarrasses you.

"Release early" is the first of Graham’s hardest lessons, and "Launch now" is the first line of YC’s pocket guide.

Install basic analytics before launch: signups, weekly active users and the one action that shows a user got value.

Seibel names "not using analytics" as a first-time founder mistake: you need to know what is being used and what is not.

Sit with three users while they use the product. Say nothing; write down every place they hesitate.

Sam Altman: "Literally watch them use your product. Sit in their office if you can."

Ship a small improvement every day or two after launch.

Graham: "You should make your system better at least in some small way every day or two."

Recruit users one at a time: write to 30 named people who have the problem and offer to set them up personally.

Graham calls recruiting users manually "the most common unscalable thing founders have to do at the start".

When someone agrees to try it, set them up on the spot: on a call, on their laptop, right then.

This is the "Collison installation" Stripe’s founders used inside YC.

Send every new user a personal thank-you, and reply to every message the same day.

Graham says to take extraordinary measures to make early users happy; Wufoo sent each new user a hand-written thank-you note.

Choose one growth number (revenue, or active users if you don’t charge) and a weekly target. Check it every Monday.

YC measures growth per week during the batch. Graham says to pick a rate you think you can hit and try to hit it every week.

Each week, rank your three most important problems, try a fix within the week, and check the result the next week.

Graham says founders should try things "at a resolution of a week or less" and report results at the next office hours.

Block at least two half-days a week with no meetings, and push calls to the end of the day.

Makers work in units of half a day, and "a single meeting can blow a whole afternoon". Graham held office hours at the end of his working day.

If you are alone, look for a cofounder: create a profile on YC’s free co-founder matching and meet three people this month.

"Single Founder" is the first of Graham’s 18 mistakes: "Starting a startup is too hard for one person." YC’s matching is free.

Hold a weekly 30-minute cofounder check-in, and write down roles and equity while things are going well.

Seibel lists not having transparent conversations with your cofounder as a top mistake; YC’s pocket guide says founder relationships matter more than you think.

Keep a setbacks log: each time something goes wrong, write what you did next. Review it monthly.

Graham: "The most important quality in a startup founder is determination. Not intelligence-- determination."

Protect sleep and exercise. Put them in your calendar like customer meetings.

The last line of YC’s pocket guide: "Get sleep and exercise – take care of yourself".

Work out if you are default alive: with expenses flat and revenue growing at its recent rate, do you reach profitability before the money runs out?

Graham says founders should know this early, and that asking too early is far less dangerous than asking too late.

Cut costs until revenue could cover the founders’ basic living expenses. Aim for ramen profitable.

Ramen profitable means you are "no longer at the mercy of investors", so you can raise on better terms or not at all.

Freeze hiring until growth is real. Do the work yourselves for now.

Graham: "Hiring too fast is by far the biggest killer of startups that raise money."

If you raise, list investors and talk to them in parallel, not one by one, and ask at every meeting what happens next.

Graham’s rules: breadth-first search weighted by expected value, and "never leave a meeting with an investor without asking what happens next".

Learn how a SAFE and the YC deal work, so you know what 7% plus an MFN SAFE would cost you.

YC’s standard deal is $500,000: $125,000 for a fixed 7% plus $375,000 on an uncapped MFN SAFE (as of November 2026).