A better question, spotted in a tweet today, separates founders who find real demand from founders who collect polite yeses.
Why does "would you pay for this?" get a yes almost every time?
Because it's free to answer. Saying yes to a hypothetical costs a stranger nothing, not money, not time, not the discomfort of disappointing you in person. A founder posted on Reddit today asking if people would pay $35 for a custom repair patch on ski gear. His own friend, someone with real skin in the friendship, laughed and said he'd just use tape. That's the most honest data point in the whole thread, and it came from the one person who wasn't performing politeness for a stranger online.
What's the difference between a signup and a customer?
A signup costs an email address. A customer costs money, and money is the one thing people are reliably reluctant to hand over for something they don't actually want. Another founder posted today about 100 signups in two weeks with zero marketing, treated by the thread as proof the idea works. It isn't proof of anything except that 100 people were curious enough to type an email into a box. None of them have been asked to pay yet. The gap between those two numbers, signups and payments, is where most startups quietly die months later.
Why did 60 cold DMs get zero replies?
A founder sending outreach for a new SaaS tool got 60 cold DMs and cold emails out, and heard back from nobody, not even a polite no. The thread debated channel (DMs vs email vs forums), messaging (leading with the product vs the pain), and trust (no testimonials yet). All reasonable. But zero out of 60 is rarely a tactics problem. If the pain were sharp enough, a bad message through the wrong channel still gets a reply, because people who are actually hurting will fight through friction to fix it.
What does a good validation question actually sound like?
Justin Welsh, an operator with a real audience, posted it plainly today: don't ask people what pays well, ask what hurts badly enough that they'll pay anything to make it go away. That single reframe changes the entire research conversation. Instead of "would you pay $10 a month for this," the question becomes "what did you pay money for last month to make a specific problem disappear." One is a hypothetical about your product. The other is a fact about their behavior, and facts about behavior are the only thing that predicts what someone will actually do next time.
Does a more natural AI voice change any of this?
No, and today's launch makes the point better than any hypothetical could. OpenAI shipped a new voice model built for more natural, real-time conversation. Within hours, reviewers were complaining it's over-enthusiastic, leaning on filler words like "mhmm" and "yeah" to sound reassuring. That's an AI optimized to feel heard, the exact same failure mode as a friend who tells you your idea is great. Sounding attentive, human or synthetic, has never once been the same as telling the truth or reaching for a wallet.
Key takeaways
- "Would you pay for this?" gets a free yes because answering it costs the respondent nothing, not money, not time, not social discomfort.
- Signups measure curiosity. Only a payment measures whether someone wanted the thing badly enough to give something up for it.
- A silent cold-outreach channel (60 messages, zero replies) usually signals a demand problem before it signals a tactics problem.
- The sharper research question isn't what someone would pay, it's what they already paid money to make go away, because that's a fact about behavior, not a hypothetical about your product.
- A more naturally agreeable AI voice model is still optimized to sound attentive, not to tell you the truth, the same failure mode as an encouraging friend.
FAQ
Why is "would you pay for this?" considered a bad validation question?
Because it asks someone to predict their own future behavior for free, and people are consistently bad at that, especially when being polite to a stranger costs them nothing.
What should founders ask instead of "would you pay for this?"
Ask what the person already spends money on to solve a specific, recent, and painful problem. Past spending on a real pain is a fact. Future willingness to pay is a guess.
Are signups a good sign of product-market fit?
Signups show curiosity, not commitment. The number that matters is how many signups convert to an actual payment or a costly action, not the raw signup count.
Does getting zero replies to cold outreach always mean the idea is bad?
Not always, but it's a stronger signal than most founders want it to be. A sharp, painful problem tends to get replies even through an imperfect channel or message.
Can a more advanced AI model help validate a startup idea?
A more advanced model can hold a more natural conversation, but conversation is still stated opinion. It cannot replace a real stranger spending money or time before your product exists.
Why this today: three separate Reddit threads (a $35 ski patch, 100 zero-revenue signups, 60 silent cold DMs) all hit the same blind spot the same day Justin Welsh posted the sharpest reframe of it on X, right as OpenAI launched a voice model already being called too agreeable.