How do you get your first users when nobody knows you exist?

You go to them. Nobody finds you at this stage. So find the room your users are already in. A niche subreddit, a Discord, a forum. Be useful in it before you ever mention your product. Supademo got their first 2,000 signups from one Reddit post. It offered to do free work for people, with no pitch attached.

Can I get users with no audience and no money?

Yes — that's the normal case. Every story in this guide started with no audience at all. Buffer's founder wrote 150 guest posts. He did it to borrow other people's readers. Airbnb went for the bloggers who had the smallest crowds. Those were the ones who replied. What it costs is your time, one message at a time.

Should I get users first or validate the idea first?

Validate first. Most guides here say you have already built the thing. They say your only problem is getting the word out. But you can recruit users for a thing nobody needs. That is a slow way to learn that nobody needed it. A live page and fifty of the right visitors will tell you in a week. No market need is the #1 reason startups fail.

When do I stop doing things that don't scale?

When the same unscalable thing has worked three times in a row. Same reason each time. That repeat is the signal. You've found a real pattern worth automating. Before that, you're not scaling a channel — you're scaling a guess.

How to Get Your First Users

Q: How many users do I need to validate a product?

Fewer than you think to learn something, more than ten to trust it. Ten users can all be friends and favours. Around fifty is where retention starts to mean a thing. Word of mouth either shows up or it doesn't. Stripe's Patrick McKenzie puts the bar well. By the time you have 10 happy customers, it isn't a fluke anymore.

In 2020, Supademo’s founders put an offer on Reddit. Drop your product URL. We’ll build you a free interactive demo. No launch, no ad budget, no audience. That post brought in their first 2,000 signups. Today the company is past $4M ARR and 200,000 users. No ads and no sales team.

That’s the shape of it. Your first users don’t arrive. You go and get them, one at a time. You get them from a place where they already are.

Most guides to this question skip a step, though, and it’s the expensive one. They all assume you’ve already built the thing. They assume your only problem is getting the word out. Sometimes that’s true. Often no one signs up because no one wanted it. No amount of clever marketing fixes that. CB Insights read the post-mortems. No market need tops the startup failure list. And we charted the top ten reasons here. Recruiting users for a product no one needs is a slower way to find out.

So this guide starts one step earlier than the others.

Step 0: find out if anyone wants it before you go hunting

Here’s the cheap version of the next six months. Put up a page that describes the thing. Send fifty of the right people to it. See what they do. If nobody signs up, you’ve learned something in a week. Otherwise it would take you a year and a product.

Rahul Vohra did roughly this. It was before Superhuman was Superhuman. He built a landing page on Squarespace in two hours. He wired it to capture emails. Then he set up an auto-reply that asked two questions. Then he wrote an article (“RIP Mailbox,” three days to write, one day to shop around) and pointed it at the page. It brought in north of 5,000 signups. Those auto-replies turned into early talks. About a thousand of them. The page came first. The product came after, shaped by what those people said.

That’s a demand test. It beats a survey for one reason. People lie in surveys. They tell the truth with their behaviour. “Would you use this?” gets you a yes. A signup form gets you a fact. A payment field gets you the truth.

Y Combinator’s advice runs the same way. Charge real money early. Not because you need the revenue. The point is the feedback. Paying customers give sharper feedback. Free users never will. A free user who drifts away loses nothing. And tells you nothing.

The full method is in the validate a startup idea guide. We line up the ones that do it in our idea validation tools comparison. Have you already got signal that people want this? Real signal, not cheers from friends? Then read on.

“First users” is three different jobs

The phrase hides three problems that need three different tactics. The three get lumped together. That’s why so much advice feels useless.

Your first 10 are a search problem, not a sales problem. Ankit Gupta at YC puts it better than we can: “Finding your first users is more of a search problem than a persuasion problem.” You’re not convincing a market. You’re finding the handful of people whose hair is on fire. Then you hand them a bucket. If you have to persuade someone hard, they’re the wrong person. Early on, a hard sell is a signal you’re in the wrong room.

Your first 50 is where you find out if it’s real. Fifty is roughly where retention starts to mean a thing. Word of mouth either shows up or it doesn’t. Ten users can all be friends and favours. Fifty can’t.

Your first 100+ is where you find the one channel that works. Then you stop doing the other twenty-eight.

Don’t skip ahead. Most founders go at the first-100 problem with three users. That’s how you end up with a marketing plan and no customers.

Pick one channel, not twenty-nine

Search this topic and you’ll get listicles with 29 numbered tactics. Ignore them; here’s the sourced reason why.

Lenny Rachitsky went and looked. How did the winning apps get their first users? The finding? Just seven strategies drove all of that early growth. And most apps used just one. Not seven. One. Those tactic-dump posts describe a menu. Nobody in the sample ordered from it.

So the job isn’t to run 29 experiments. It’s to guess which single room your users are already in. Go there. Be useful for a month. The real channel set is short:

Communities where they already gather. A niche subreddit, a Discord, a Slack, a forum. Highest hit rate for most software.
Answering people who already asked. Public questions in those same places, replied to one at a time, by hand.
A borrowed audience. Someone else’s newsletter, podcast, or fans.
Launch platforms. Product Hunt, Indie Hackers, BetaList. A spike, not a channel.
Your warm network. The first five, and only the first five.

Pick the one where your users are densest. Not the one you’re most comfortable with. That last part is where founders quietly cheat.

Communities: go where they gripe

This is the best channel for most early products. It’s also the one people do worst. They show up and pitch.

Supademo’s Reddit post worked. It wasn’t a pitch. It was an offer: give me your URL and I’ll do the work for you, free. The value came first, the product came second, and 2,000 people took them up on it.

There’s a reason to care about this beyond signups. Reddit is now the #1 cited domain in AI answers. It shows up in roughly 21% of AI Overviews. And you can watch it happen on this very topic. Google’s AI Overview for “how to get your first users” cites r/SaaS. Being useful in the right subreddit pays off. It didn’t three years ago.

The method matters enough that we gave it its own guide. It’s how to get users from Reddit. It has the value-first post template. It also has the 90/10 rule that keeps you from getting banned in week one.

Answering people who already asked

The highest-yield version of this is not a message to a stranger. It is a reply to somebody who has already posted the problem in public and asked what to do about it. They have raised their hand; you are answering, in the open, where the next hundred readers see it too.

Here is the shape that works. Steal it:

[The specific thing they described, said back to them in one line, so it is
obvious you read the post.]

I ran into this too and ended up building [thing]. [One sentence on what it
does.] Not selling anything here — happy to just tell you how I'd solve it
without it.

Why it works: it proves you read the post, it is specific, and it asks for judgement rather than money. Write thirty of these by hand. Don’t automate them, and don’t move them into somebody’s inbox — the automation is what kills the reply rate, and the inbox is what gets you reported.

Zapier’s first paying customer came out of a forum post. Wade Foster answered a question, built the exact integration the person needed, and closed them. That’s not a growth hack; it’s a founder doing a job by hand.

Borrowed audiences: someone already has your users

You don’t have an audience. Someone else does.

Buffer is the clean case. Leo Widrich wrote 150 guest posts in the company’s first nine months. That brought in roughly 100,000 users. Content was driving 70%+ of daily signups. That’s a grind with no glamour. It worked because he went to where the readers already were. He didn’t wait for them to find Buffer’s blog.

The same lever, smaller. Airbnb’s founders went after bloggers. They went for the ones with the smallest audiences of all. Those bloggers actually replied. Twitter got 250+ signups the day after Om Malik wrote about it. Fewer than 600 people were on the service at all.

Do things that don’t scale — and know when to stop

Paul Graham’s line is the most-quoted advice here. It’s also the least finished. Everyone tells you to do things that don’t scale. Nobody tells you when to stop.

The manual playbook is real and it works:

Facebook launched to the Kirkland House mailing list — about 300 people. That produced 1,200–1,500 registrants in 24 hours.
Slack hand-recruited teams it knew, and got 8,000 invite requests on day one and 15,000 within two weeks.
DoorDash put flyers up around Stanford. Tinder threw parties at USC sororities. Etsy went to craft fairs. Nextdoor worked HOA boards.

Patrick McKenzie’s rule from Stripe Atlas is the one to learn. The founders make the first sales themselves. Not a hire, not a funnel. You. And his payoff line is the useful bar. By the time you have 10 happy customers, it isn’t a fluke anymore.

So when do you stop? Here’s an answer, since nobody else gives one. Stop when the same unscalable thing has worked three times in a row. Same reason each time. That repeat is the signal. You’ve found a pattern you can build a process on. Before that, you’re not automating a channel. You’re automating a guess.

Doing things that don’t scale isn’t a phase you suffer through. It’s how you learn what the scalable version should be. Vohra’s thousand talks weren’t a bootstrapping tax. They were the product spec.

What not to do yet

Paid ads. You don’t know your message, your audience, or your conversion rate. Ads will convert your ignorance into a bill.
SEO. It works (it’s most of why you’re reading this), but it pays out in months. It’s not a first-user channel.
Automating any of it. You write thirty replies by hand, so there is nothing to automate. Automating them is what tanks the reply rate, and what gets you banned.
Hiring a salesperson. If the founder can’t sell it, a hire can’t either. They’ll just fail more expensively.
A second product. The temptation when the first one is quiet. It’s almost always avoidance.

The ladder

Put together, the sequence has gates. Don’t climb until each one is true:

Demand test. A live page, real traffic, real signups or payment intent. Gate: strangers act, not friends.
First 10. Hand-recruited from one community, by you. Gate: they use it unprompted, twice.
First 50. Same channel, repeated. Gate: retention holds and someone refers you without being asked.
First 100. Now, and only now, work out what scales.

Most founders start at step 3, wonder why it’s not working, and blame the channel.

Before you go hunting for users, find out if they’re out there. ProofMachine digs out the word-for-word complaints and names the exact communities they came from. That’s where you recruit your first ten.