All hubs

Idea

Finding a startup idea worth working on

Most public startup-idea advice is bad: "solve a problem you have," "look for big markets," "find white space." Useful only if you already know what good looks like. This hub distils what experienced founders actually do — find ideas that pass the filters, kill ideas that don't, and avoid the seductive ones that look brilliant on a Friday night and rot by Monday.

Last updated May 21, 2026

Who this is for

Aspiring founders deciding what to build, and existing founders questioning whether their current idea is the right one.

What you'll learn

  • Where good startup ideas actually come from
  • How to spot ideas that look big but are bad
  • The 5-filter test for any idea you're considering
  • What to do when you have too many ideas (or none)
  • How long to spend on an idea before committing
Score your idea now

Where good ideas actually come from

Three sources, in this order:

  1. Acute personal pain — you tried to do a thing, couldn't, would have paid to. The friction was real, not theoretical. Bonus: you can describe the customer in one sentence (it's you).
  2. Watching one customer for an hour — sit next to a friendly stranger doing their job. The thing they swear at five times is the idea. Most founders skip this; it's the highest-density idea source on Earth.
  3. A specific narrow wedge in a giant market — not "fix healthcare", but "the form that physical therapists fax to insurance every Tuesday."

Sources that produce mostly bad ideas:

  • "What's the next big thing in AI / crypto / VR" — thematic hunting without a customer.
  • "I read this trend report" — by the time it's a report, the wedge is gone.
  • "ChatGPT gave me five ideas" — none of them have a customer.

The 5-filter test

Run any idea through these. An idea that fails one filter is not necessarily dead, but you need a real answer for it.

  1. Who specifically pays? Name the role + company size + situation. "SMB owners" is not specific.
  2. What do they pay today (in money or hours)? Real budget proves the problem exists.
  3. Why now? What's changed in the last 12-24 months that makes this possible? If nothing has, ask why no one's solved it already.
  4. Why you? Unfair advantage = experience, network, distribution, or sheer willingness to suffer. Founders without one usually quit at month 9.
  5. What's the first 100 customers playbook? If you can't name a channel that gets you 10 customers in 30 days, the idea may be valid but the wedge isn't.

Too many ideas vs no ideas

If you have too many: This is a discipline problem, not an idea problem. Pick the one with the strongest filter score and commit for 90 days. Track new ideas in a parking lot doc; don't let them derail the current bet.

If you have none: You haven't watched enough customers do their job. Block five conversations this week with people doing work you find interesting. Don't pitch — listen. Take notes on the exact moment they swear or sigh. That's the idea source.

The trap in the middle: "I have a vague intuition but no specific wedge." Vague intuitions don't survive contact with customers. Sharpen by talking to 5-10 candidate customers; the wedge appears in their language, not your slides.

Step-by-step action plan

Do these, in order

  1. 1Block 5 customer-discovery conversations this week — don't pitch, just watch them work
  2. 2Write down any idea that passes the 5-filter test; reject the rest
  3. 3For your top idea, write the first 100 customers playbook in one page
  4. 4Talk to 10 candidate customers and confirm the problem in their words, not yours
  5. 5Commit to one idea for 90 days or 50 conversations, whichever is first

Frequently asked questions

Should I quit my job before I have a validated idea?
Almost never. Validate in nights and weekends until you have either signed revenue or 10+ explicit customer requests. Quitting before that creates pressure that produces bad decisions.
Is it OK to copy an existing idea?
Yes, especially in B2B SaaS. Most successful companies are the second or third entrant in their category. The angle to be different on is the customer segment, the distribution, or the wedge — not the entire concept.
How long should I spend on a single idea before pivoting?
90 days of focused work or 50 customer conversations, whichever comes first. Before that, you don't know enough to pivot. After that, if no real demand signals, the idea probably needs to change.
Are AI / crypto / VR ideas overhyped?
Themes are overhyped; specific wedges aren't. "Build something with AI" is a bad idea. "Use AI to automate the back-office task that physical therapists do every Tuesday" might be a great one. Theme is the layer of the stack; idea is the wedge.

Related resources

Related tools

Related templates

Related courses

Related hubs