How to Get Your First Users With No Marketing Budget

With AI, building is easy, so you built a tool to solve a problem. Maybe you have a landing page, a few GitHub stars, or a launch post (that may or may not have been favorably received). And there are close to zero users, if not zero. This is where a lot of technical founders land right now, and it's where you run into the thing everyone kept telling you, and you didn't quite believe: building was the easy part, and distribution is the hard part.

I've been building SaaS products solo for two years, and I answer some version of these questions on Reddit almost every day:

• How did you get your first users?
• I skipped distribution, and I'm stuck at zero.
• Sixty-six stars, no users - what am I missing?

My answer varies depending on the starting point, but it centers on the same few themes: go to where your problem is already being discussed, be useful there, and let a funnel start. That's the whole method; the rest of this post explains how to run it.

Is finding users actually the hardest part of building a SaaS?

For most technical founders, yes, and the reason is buried in how you got good at building. Your whole career has rewarded explaining how things work in precise, domain-specific language. Distribution asks for the opposite.

In order to do distribution well, you first need to learn how your ICP (ideal customer profile) frames the problem in their own words, and, when it fits, your job is to connect your tool to their framing (not yours). The reason the community-driven method (see below) works is that it forces you to learn your ICP's framing in the first place, then meet them where they are to provide help.

It also explains the pattern you keep seeing. Ad clicks, likes on social media posts, and landing-page visits that never convert are attention, not adoption. You reached people who appreciated the concept, not people who have the problem. The fix is to stop pushing your tool and instead pull in people dealing with your problem.

How to validate a SaaS idea and get your first users (on Reddit or anywhere)

Here's the process I use, and it works with zero marketing budget.

1. Find the communities where your ICP already gathers. Reddit, Hacker News, X, Slack groups, Discord servers, wherever your ICP actually spends time.
2. Confirm the problem is real by reading, not posting. Count how often your problem comes up on its own (in threads you had nothing to do with). What you want is for other people to raise the problem organically. If the same pain shows up week after week in language you recognize, that's a validated problem. If you have to look hard, or it barely surfaces, that's an answer too, and it's cheaper to learn that you have the wrong problem or the wrong ICP now rather than after spending more time building.
3. Join the recurring discussion and actually help. Offer something useful in the thread: what to look for, a manual way to make progress today, or a mistake to avoid. Give real advice with no strings. As a builder, the trap is over-indexing on your product and under-indexing on the person. Being useful means helping them get unstuck right now, even when the manual fix you describe is exactly what your tool automates. You're being helpful whether or not the person ever becomes a user, and that's the point.
4. If you have an MVP, share it where it fits. Alongside the advice, when your tool genuinely answers their question, share a link to your landing page. The link is a byproduct of being useful, dropped only when it actually helps the person asking the question.

Do this, and you'll start to see a trickle of activity from people who actually have the pain you built for. When those specific people sign up, download, or use the thing, you have the start of a real marketing funnel.

Outbound vs inbound: which gets your first users?

This method sits closer to inbound because you go where demand already exists rather than creating it from scratch. At zero brand, you need to sit as close as possible to the people experiencing your problem. The work feels inefficient, since you're only engaging where the problem is already being discussed, one thread at a time. But engaging with your ICP while their pain is live is how you learn what actually lands. It's also your sharpest early signal. If people actively experiencing your problem won't use your product, the problem is either your product or your framing, and the approach needs to change.

Outbound and scaling come after this. Without proof of a problem, a message that resonates, or some usage, there's too much uncertainty, and you risk wasting time, your existing network, and money trying to reach early users. Once you've done this high-touch work long enough, you start to see the message that reliably turns a stranger into a user. That's the thing you scale or use in cold outreach.

Turn the conversations into compounding SEO

The community work is high-touch, but it feeds something that compounds. Pay attention to the language people use, especially the top-level question that sparked the discussion and how others frame their advice.

Two things to do with that language. First, put it into your website copy. Second, write posts that answer the common questions so you can explain your thinking in depth. If people ask a question in a community, they're also asking AI or a search engine, so your site should answer in the same words. This means the high-touch work in communities slowly turns into passive inbound that arrives while you sleep.

The tedious part is finding the threads

The honest weak point of this method is steps one and two. Finding the communities and reading them closely enough to catch the handful of relevant threads eats into your day. You end up context-switching every couple of hours to scan for the few posts where you actually have something to add.

I open-sourced OpenMagpie to help other builders solve the scanning problem. You point it at your communities, describe in plain English what a relevant post looks like, and OpenMagpie surfaces the threads worth showing up in so you can spend your time helping instead of searching. It runs on whatever LLM you point it at, including a small local model like qwen3:8b (what I run), so it costs close to nothing to get started.

Start this week

This is a high-touch process that depends on being genuinely useful, which is exactly why it works. If you're having trouble reaching initial users, you don't need to watch your posts get buried, send 500 messages into the void, or spend on ads. Pick one community this week, find the recurring version of your problem, and help in one thread. That's day one of zero-spend marketing, and it's the same day one I keep coming back to myself.