who is this for?

A specific person with a specific, painful problem beats a large vague market every time at the start. The narrower and more real your first user is, the faster everything else gets — the product, the pitch, the sale.

The tell that you’ve gone too wide is that your product starts to look like anything, and anything looks like Claude. When a VC asks “what if Claude does this” or “what if Google does this”, the real message is that your ICP is so broad the thing has no shape. A founder told me his ICP was “creators and B2B”. That’s everyone, which means no one. The fix is to narrow until you can picture one person: traders who’ve traded a hundred times this month, one to five million in volume, already running a script and wanting to move to AI. That specific. Now there’s a face to the problem.

Go for the few who already feel the pain most. The fifty or a hundred power users who do the thing manually every single day, not the thousand who’d find it mildly nice. One whale with two million is worth thirty retail users with five hundred dollars each, and he introduces you to his friends. When you ask your network for help, don’t ask for names — ask for a number. “Anyone above 500k, connect me.” A list of names is noise; a number is an ICP.

And pick a customer you can actually reach. Don’t go to family — anyone who knows you takes you for granted and says nice things. Go to a similar business where a thirty-year-old just took over from their father; they’re easier to convince than the sixty-year-old, and once you’ve done it for one, the whole industry can see it.

Write your ICP as one sentence describing a single person with a number in it. If the sentence needs an “and”, you’re still hedging two audiences. Cut until it’s one.