Most founder advice about "finding ideas" is bad because it treats ideas as problems to solve by brainstorming. They're not. Good startup ideas are problems you noticed.
Paul Graham frames it as: the way to get startup ideas is not to try to think of startup ideas. It's to look for problems, preferably problems you have yourself. The strongest filter is asking yourself "Why doesn't this thing exist?" — and finding that the answer is interesting.
Three filters for a real idea
A real startup idea passes three checks:
- Someone you know has the problem. Not "the world." A specific person.
- The problem is annoying enough that they're already kludging a fix. Spreadsheets, manual workflows, copy-pasting between tools.
- You're somehow well-positioned to solve it. Domain knowledge, technical skill, or weird-life-experience that gives you an advantage.
If you can't say "I'm building this for [specific person]" — you're brainstorming, not idea-finding.
What "schlep blindness" is
Some of the best ideas exist in plain sight but feel unattractive. The work looks tedious. Stripe was a card-payments API; the work involved was banking integrations and compliance — boring schleps. PG calls this schlep blindness: the mental flinch that hides good ideas because they look like work.
Tell yourself the truth about market size
A small market that desperately wants the thing is infinitely better than a huge market that's lukewarm. "Make something a small group of people love" beats "make something a large group of people kinda need."
This sounds counterintuitive but matches what every successful early-stage founder will tell you. Airbnb started with conference attendees who couldn't find hotel rooms in San Francisco. Stripe started with the founders' own friends shipping side projects. Tiny.
If you can't fill that line in honestly, you don't have an idea yet. You have a topic.
What to do this week
- List 5 problems you encountered in the last month that frustrated you.
- For each, write down: who else has this problem, what do they do today, and what would the dream version look like?
- Pick the one with the most desperate user. Not the biggest market.
That's your starting point.