Since building Needle, we've talked to hundreds of founders and learned a lot about what actually works when it comes to finding your first customers. Most advice out there is either too generic or aimed at companies that already have traction. This post is different: it's specific, honest, and bucketed into pre- and post-product-market fit. We've also included a couple of areas we haven't nailed yet - we're still learning.
One more thing: we built Needle because we kept hearing the same problem. Founders were either manually checking Reddit, Hacker News, Twitter, and a dozen other places (and burning hours), or they were guessing where their customers hung out. So when we say "use a tool" or "see where conversations actually happen," we're not being coy - we use Needle for this ourselves and built it for exactly that. With that said, the principles below stand on their own.
Pre–product–market fit
Compete on depth, not breadth
Trying to be everywhere at once is a losing game when you're early. You're not a big brand with a content team. So focus: one or two channels done well beat ten done poorly. The catch is knowing which one or two. That's where evidence beats gut. Run a few searches (or use a tool like Needle to search across Reddit, Hacker News, Stack Overflow, Twitter, and more in one go) and see where your ideal customers actually discuss the problem you solve. You'll often find that one platform has way more relevant, high-intent conversations than the others. Double down there before spreading thin.
In practice: One deep channel where people are actively asking for solutions beats a scattergun across five. Quality and focus beat quantity every time.
Ask: "Is this genuinely useful to someone looking for a solution?"
This applies to every piece of content and every outreach message. If you wouldn't be proud to share it on your own feed, don't publish it. SEO posts aren't exempt - "it's for keywords" is a bad excuse for thin content. The same goes for how you show up in conversations: be useful first. That's the heart of early adopter outreach - find people who are already looking, then help before you sell.
Treat your website like a product
Your site is often the first real touchpoint. Keep it clear, fast, and honest. Resist the urge to make it "marketing-y." Developers and thoughtful founders can smell fluff. We try to keep our own site straightforward: what we do, who it's for, how it helps. No dark patterns, no fake urgency. It's the opposite of most SaaS landing pages - and that's on purpose.
Be really careful about outsourcing customer discovery
Outsource the things you genuinely can't do, not the things you're too busy to do. If you outsource "find our customers" or "figure out where they hang out," you'll spend more time briefing and correcting than if you'd just done it yourself. Use tools to make the job faster (search across platforms, use brands (Auto) and digests so you are alerted when your brand or space shows up in new threads), but keep the understanding in-house. Once you know where your people are and what they care about, you can outsource execution - content, ads, design - with a clear brief.
Hacker News (and Reddit, and Twitter) are double-edged swords
A front-page hit feels amazing and can drive a spike in signups. But it's not a strategy. Even if you're good at it, you'll have a low hit rate. The founders we see win long-term are the ones who use HN, Reddit, or Twitter as one channel in a repeatable system - not as a lottery ticket. Before you bet big on one platform, check where the most relevant, ongoing conversations actually live. Often it's not the platform that gives the biggest ego boost. Tools that let you search across multiple platforms at once can surface that quickly so you don't waste months on the wrong place.
No secret tricks. Don't ask friends to upvote. Don't game the system. Post good work, be helpful in comments, and build a repeatable engine so you're not dependent on one viral moment.
Beware the attribution mirage
UTM parameters and last-click analytics will lie to you. Someone reads about you on Hacker News, then later searches your name and clicks a Google ad. Your dashboard says "Google Ads are crushing it!" - but the real source was HN. In your signup or onboarding flow, add an optional free-text field: "Where did you first hear about us?" Only a fraction will fill it in, but that data is gold. Review it regularly. Cross-check it with where you're actually seeing the best conversations (e.g. which platforms have the most relevant threads about your space). That qualitative picture is often closer to reality than any attribution report.
Post–product–market fit
Keep someone close to the customer on the team
Once you have traction, it's tempting to hand "marketing" or "growth" to someone who lives in campaigns and funnels. Don't let customer discovery go dark. Have at least one person (often a founder) who still runs searches, reads real conversations, and uses Search plus trending problems to stay grounded. That keeps you from drifting into "we already know our users" and missing a shift in how people talk about the problem or your category.
Paid ads: learn where your audience actually is first
Paid can work, but it's easy to burn a lot of money on the wrong channel. Before you pour budget into Google, LinkedIn, or Twitter, get evidence on where your ideal customers actually discuss the problem you solve. Run multi-platform searches, look at trending problems in your space, and see which communities have the most intent. Then invest in paid there. We've seen founders assume "our buyers are on LinkedIn" when the best conversations were on Reddit and Hacker News - and vice versa. A few hours of discovery can save months of wasted ad spend.
Default ad settings are dangerous. Never use out-of-the-box targeting on Google, LinkedIn, or Twitter. And if you sell something over ~$10/month, be cautious with Product Hunt–style launches: you'll get signups, but many of them hate paying. Know your channel before you spend.
It's okay to experiment with sponsorships
Sponsorships (newsletters, podcasts, events) are expensive and hit-or-miss - unfortunately you often have to spend to learn. Newsletters are a good place to start. Ask your users and your team what they actually read, then sponsor those. Do bursts (e.g. three months on, three off) and rotate copy so you don't exhaust the audience. Use what you've learned from search and customer discovery to pick newsletters and communities that match where your ICP already spends time.
Two things we haven't figured out yet
Events are 10x more work than they look
We don't host our own - we don't have the budget to do it well. We do attend others when it makes sense. If you have someone on the team who genuinely enjoys events and can speak credibly (often an engineer or product person), that's more valuable than sending only "marketing" people. Sponsoring big events is often disproportionately expensive; if your logo is next to Google's, you're probably overpaying. For now we're cautious and selective.
Social media: we're still learning
We're seeing promising results in some channels, but it's too early to share a playbook. One lesson: don't try to win every channel at once. You'll feel like you did an "okay" job everywhere and never learn what actually works. Pick one, run it properly for a while, then decide. And the teams that make social look effortless usually put a lot of work in - don't underestimate the effort.
What next?
If you're pre-PMF, start with validating your startup idea using real conversations, then find your first 100 customers with a system, not a hope. If you're past PMF, keep discovery alive - use multi-platform search and Auto Search + digests (one brand profile on every paid plan) so you're not flying blind. Needle focuses on search across communities, Auto Search, and trending problems so you stay close to how people talk about the problem. And whatever stage you're at: compete on depth, be genuinely useful, and don't trust last-click attribution. We're still learning too. If you've learned something that contradicts or refines this, we'd love to hear it.