It’s a Saturday night in July 2026. You’ve got a landing page, a half-built prototype, and a hunch.
What you don’t have is a single person who’s actually confirmed the hunch is real.
So you do what every “validate your idea” post from 2023 told you to do. Build a list. Sixty LinkedIn profiles who look like your buyer. Write each one something personal. “Quick 15 minutes, I promise I’m not selling anything.” Hit send, sixty times.
By Tuesday: two replies. One books a call, then goes quiet forever.
That’s not a weak subject line. That’s not bad luck either.
The playbook died. Nobody sent you the memo.
How do you find people to interview when validating an idea?
Find people to interview by going where they’re already complaining about the problem, not by messaging a cold list you built yourself. Search niche communities, forums, and review sites for people describing your exact pain, then ask for five minutes of their expertise, framed as research, never a pitch.
People who ignore a form will still say yes to a real, curious human.
That’s the whole game. Everything below is how you actually run it.
Most advice hands you a longer contact list. That’s the trap.
You don’t have an outreach problem. You have a location problem.
Why your cold DM list is dying in the inbox
Cold outreach isn’t hard because you’re bad at writing messages.
It’s hard because the whole channel quietly stopped working over the last three years, and most advice still hasn’t caught up.
Start with email. An analysis of 53 million cold emails sent between January and June 2026 found that the average reply rate sits at 3.7%, and only 3 to 5% of replies come from a genuinely interested person rather than a bounce, an auto-reply, or a polite no.
Send sixty emails. Expect two or three real answers, if you’re lucky.
LinkedIn isn’t much better. The average cold connection request gets accepted around 29.6% of the time, and a request with zero shared context or mutual connection falls closer to 2 to 5%.
You’re not being ignored. You’re being filtered.
AI now reads your message before a human ever does.
Inbox tools built to catch personalized, AI-written outreach ship as a standard feature in 2026, not a niche add-on. The polished, custom-feeling message that used to work is now the exact pattern they’re trained to catch.
And the second front closed at the same time. Communities got strict.
Karma minimums. Account-age gates. Threads that exist purely to keep pitches out of the real conversation. Community platforms like these run on earned trust, and they reward authenticity while quietly punishing anything that reads as promotional, whichever forum it’s on.
Two doors just closed. Here’s the one that’s still wide open.
Go where the complaint already exists
Every group of people with a real problem is already describing it somewhere. Out loud. Unprompted. For free.
Subreddits built around your buyer’s job. Niche Slack and Discord communities. Review pages for the tool they’ve half-adopted and resent. The comment section under a LinkedIn post that named their exact frustration.
Somewhere right now, your future customer is typing out the problem you’re trying to solve, in their own words, to nobody in particular.
That’s your list. Not a spreadsheet built from a lookalike audience.
One founder trying to validate a B2B idea said it plainly, inside a startup community with a strict no-pitching rule: a landing page with a signup form never felt right for B2B. Busy professionals won’t fill out a random form. But they’ll hop on a call if you’re genuinely trying to solve their problem.
Sit with that for a second. The form is the ask nobody wants to do. The conversation is the one they’ll actually give you.
Lurk a community for a week before you ask anyone for their time. Learn who answers questions, who’s flagged the same pain twice, and what the moderators actually enforce. You’ll find better prospects, and you won’t get banned doing it.
Once you know where the room is, you still need to get invited in.
Ask for five minutes of research, not a pitch
Here’s the line that changes everything: you’re not selling anything. You’re asking someone to teach you something they already know.
One founder who’d run this exact playbook boiled it down to four moves: find the person’s email or LinkedIn, write something personal and short, tell them you’re doing market research, and ask if they have five minutes for a couple of questions. Don’t be salesy. It’s fine if they say no.
That structure works because of something psychologists mapped out sixty years ago. Researchers Jonathan Freedman and Scott Fraser found that people who agree to a small, easy request become far more likely to agree to a bigger one later, a pattern now known as the foot-in-the-door effect.
A five-minute ask is small enough to say yes to without thinking. “Can I pick your brain about my startup” is not.
- Lead with their expertise, not your idea. “I’m trying to understand X, and you clearly deal with it” beats any version of “I built a thing.”
- Name the time, small. Five or ten minutes reads as a favor. Thirty reads as a meeting they need to prep for.
- Never mention your product first. The second you pitch, they switch into polite mode, and you lose the honest answer you came for.
“Hi [name], I saw you deal with [specific problem] at [company or context]. I’m researching this space and would love five minutes to understand how you handle it today. Not selling anything, just trying to learn from someone who lives it.”
Once you’ve found the room and written the ask, the last problem is just arithmetic: whose problem is worth your Saturday?
Start with the people one hop from you, then widen
Before you message a single stranger, burn through the free layer. Past coworkers. Past customers from a previous job. People who already follow you because you built something adjacent.
That gets you three or four conversations without asking a stranger for anything.
Then widen to the second hop. Ask each of those people for one name. A referred intro converts at a completely different rate than a cold one, because the trust already transferred before you said a word.
Only once both layers run dry do you go looking in public rooms, and by then you already know exactly who you’re looking for. If you haven’t nailed that down yet, define your ICP before you start recruiting so you’re chasing a type of person, not a vague vibe.
Peers doing you a favor rarely expect anything. For anyone senior or outside your network, a gift card in the $50 to $150 range is a common thank-you, and it noticeably raises your reply rate without turning the conversation into a paid focus group.
Keep going until the same problem shows up three times, unprompted. Ten to fifteen real conversations is usually where a genuine pattern stops looking like a coincidence.
What to actually ask once they say yes
You got the yes. Don’t waste it by pitching anyway.
Ask about their life, not your idea. Dig into specifics from the past, not hypotheticals about the future. Let them talk far more than you do.
We’ve written up the exact questions and traps to avoid the moment the call starts, and it’s the difference between five wasted minutes and five that actually validate something.
Once you have a real signal and want to test whether people will act, not just talk, that’s a different tool: run a fake door test to measure actual demand once you already know what to build.
The first ten to fifteen conversations, do yourself. You need the pattern-matching more than you need any software.
But once you know the right questions and you’re trying to run them past thirty or forty more people than your calendar has room for, that’s a scale problem.
hollie, holito’s AI agent, can have that same five-minute conversation with each person on their own channel, then hand you back what they actually said, ranked by how often it repeated.
Frequently asked questions
How many people should I interview to validate a startup idea?
Ten to fifteen real conversations is usually enough to see a genuine pattern, as long as you’re finding people who actually have the problem. Five conversations with the right audience teach you more than fifty with the wrong one.
Should I pay people for a customer discovery conversation?
Peers and warm intros usually do it as a favor. For anyone senior or outside your network, a gift card in the $50 to $150 range is a common thank-you, and it raises your reply rate without turning the conversation into a paid survey.
Where do B2B founders find people to interview with no existing audience?
Start one hop from your own network, then widen into the niche communities, review sites, and comment sections where your buyer already complains about the problem. Recruit with a small, specific ask for their expertise, never a pitch for your idea.
Is a landing page with a signup form a good way to find interview candidates?
Not for B2B recruiting. Busy professionals skip forms but will take a short call from someone genuinely trying to understand their problem. Save the landing page for testing demand once you already know what you’re building.
Stop building the list
The people who can validate your idea are already talking about the problem you want to solve.
You just haven’t gone to listen yet.
The Bottom Line
Cold DMs are dying because AI filters catch them and communities ban them. Go where the pain already lives, and start one hop from your own network before you message a stranger.
Ask for five minutes of someone’s expertise, never a pitch. Once you’re past the first fifteen and need to run this at scale, hollie can have those conversations for you and bring back what people actually said. Try holito.
