How to Get Your First 100 Customers

Here is the conclusion first, because everything else flows from it: your first 100 customers will come from doing things that do not scale, and that is correct. You are not building a system yet. You are finding a product. The tactics that work at 100,000 customers — SEO, paid acquisition, virality, category dominance — require either money you don’t have or social proof you haven’t earned yet. The tactics that work at 0 customers are uncomfortable, personal, and slow by design. Embrace that. The founders who try to skip the unscalable phase never get to the scale phase.

Here is how to actually do it.

Why Scalable Tactics Fail at Zero

Every channel that works at scale depends on existing evidence. SEO needs backlinks and indexed pages and domain age. Paid ads need a conversion rate to optimize against, and you need at least 50 conversions before most ad platforms can learn anything useful. Word of mouth needs customers who are successful enough to talk. Product Hunt and Hacker News launches need a product polished enough not to get ripped apart in the comments. Influencer partnerships need a product credible enough for someone to stake their reputation on.

At zero customers you have none of these. Running Google Ads at zero customers typically means paying $10–$50 per click to send strangers to a page where nothing is proven, the copy is generic, and they bounce. You will burn money at conversion rates below 0.5% and learn almost nothing useful.

The alternative is not to acquire customers slowly. It is to acquire them through methods where you have advantages that don’t require scale: personal trust, direct conversations, and the ability to move fast on feedback. Those advantages disappear as you grow. Use them now.

Realistic baseline: if you’re an AI product with a clear ICP and a working MVP, a focused personal network + cold outreach + community effort over 60–90 days should get you to 20–40 customers. Getting from 40 to 100 requires those early customers to be genuinely successful, which generates the word of mouth and case studies that let you try the slightly-more-scalable channels.

Your Personal Network Is Customer Source #1

This is the most reliable channel you have and most founders underuse it because it feels embarrassing. It isn’t. You are not asking for charity. You are giving someone early access to a tool that solves a real problem for a price that reflects the product’s current state. That is a fair exchange.

Work through your network systematically. Go through your LinkedIn connections, your email sent folder, your phone contacts, and your Slack DMs. Filter for people who work in the industry you’re targeting. Make a list. Be specific: not “people in tech” but “people who manage engineering teams” or “people who work in compliance at financial services firms.” Aim for 50–100 names.

Then write to each one personally. Not a bulk email. Not a newsletter blast. A one-paragraph message that names what you built, why it’s relevant to them specifically, and asks for 20 minutes or a free trial, not a sale.

What does not annoying look like in practice:

Hey [name] — I built something you might actually want to use. It’s a tool that [one sentence on what it does] — I built it partly because I kept hearing [their industry] people complain about [specific pain]. I’d love to get you into the beta and hear whether it resonates. No pitch, just a free trial and 20 minutes on the phone if you have time. Want me to set you up?

Do not mass-send this. Do not automate it. Write each one. Reference something real about the person. A 30% response rate on 80 personal messages is 24 conversations. If you can convert 40% of those to active trials and 30% of trials to paid, that’s 3 customers from a single afternoon of writing. Do that 10 times over 3 weeks and you’re at 30 customers.

The common mistake: people send five messages, get two responses, close one deal, and conclude the personal network is exhausted. It isn’t. You’ve messaged 5 out of 80 people and you stopped because the rejection stings more than the outreach is worth to you psychologically. That’s the real problem, not the channel.

Cold Outreach That Actually Works for AI Products

Cold outreach is not spam. Spam is untargeted, generic, and sent at volume. Cold outreach that converts is hyper-specific, relevant to the recipient’s actual situation, and sent to a small enough list that you can personalize every message.

For AI products, the most effective cold outreach names the specific inefficiency your product fixes and connects it to something observable about the recipient’s business. This takes 5 minutes of research per contact but typically triples reply rates.

What to research per contact:

• Recent content they’ve posted or published that relates to your product’s domain
• Job title and what their role likely involves day-to-day
• Their company’s tech stack or workflows if public (job postings are a gold mine)
• Any shared context — mutual connections, conferences, communities

Subject lines that get opened:

• “[Their company name] + [specific problem your tool fixes]”
• “Quick question about how [their role] handles [specific workflow]”
• “Saw your post on [topic] — built something relevant”

Subject lines that don’t:

• “AI tool for [generic category]”
• “Introducing [your product name]”
• “I’d love to learn more about your work”

The body should be three short paragraphs: what you noticed about their situation, what you built and what it does in one sentence, and a single specific ask (usually a 15-minute call or a free trial setup). No more. Long emails get skimmed and ignored.

Realistic numbers: a well-targeted list of 200 contacts with strong personalization should yield a 15–25% reply rate and a 30–40% conversion from reply to call. That’s 10–20 calls from 200 outreach messages. At 50% close rate on calls, that’s 5–10 customers. Not spectacular — but it’s predictable and repeatable.

Community-First: Join Before You Sell

Every ICP has communities — Slack groups, Discord servers, Reddit forums, LinkedIn groups, niche newsletters with comment sections, in-person meetups, and conferences. You need to be in those communities before you try to sell anything.

The pattern that works: join the community, spend 2–4 weeks being genuinely useful (answer questions, share resources, engage with posts), and only then mention what you’re building — and only when it’s directly relevant to a conversation that already exists. This works because trust in communities is built through demonstrated value, not announced value.

The pattern that gets you banned or ignored: join the community, immediately post “check out my new AI tool,” include a referral link. Every niche community has seen this thousands of times. It generates resentment, not signups.

Finding the right communities: search Reddit for subreddits related to your ICP’s job function (not their industry — their actual problems). Look for Slack communities in your niche (many are listed on slofiles.com or community-built lists). Look for active LinkedIn groups. Look for Discord servers attached to popular tools in your space.

One useful tactic: when you answer questions in these communities that relate to your product’s domain, you naturally build a reputation. When you eventually mention what you’re building — “I’ve been working on a tool that handles exactly this problem, happy to share access if anyone wants to try it” — it lands differently than cold promotion. Conversion rates on this kind of mention are dramatically higher than cold outreach.

The Concierge Approach: Manual First

The fastest way to get someone to pay for a software product they haven’t seen before is to do the work for them first and show them the result. This is the concierge approach: you deliver the value manually, prove the outcome, and then explain that your product automates it.

Concierge selling for an AI product looks like: taking a potential customer’s real data or real use case, running it through your tool (or doing the work by hand if the tool isn’t ready), and handing them back a polished output. They experience the value before they’ve bought anything. The question shifts from “will this work for me?” to “how much does it cost to get this automatically?”

This approach is underused because it’s time-intensive and doesn’t scale. That’s the point. You are not trying to scale. You are trying to close 10 customers and learn what they actually care about. Spending 2 hours per customer to close the first 10 is a rational trade. Those 20 hours will teach you more than any amount of analytics.

The conversion script for a concierge close: after you’ve delivered the manual output, ask directly — “Does this solve the problem you described? If I can automate this so it runs on your own data in under 30 seconds, what would that be worth to you monthly?” Let them anchor the price. Then close at or below their number.

Partnerships and Co-Selling With Complementary Tools

Most early-stage founders think about partnerships as something to pursue after they have traction. The reality is that partnerships are often the fastest path to the first 20–30 customers when your direct channels are still cold.

The logic: a complementary tool already has customers in your ICP. If they recommend you to their user base, you get warm leads with built-in social proof. The bar for a partnership is lower than you think — you don’t need to be equals. You need to be useful to their customers and not competitive with their product.

Finding partners: list the tools your ideal customer already uses. Look for products that are adjacent to your workflow but don’t solve your specific problem. Reach out to the founder or head of partnerships directly. The pitch is simple: “Your customers deal with [problem]. I built a tool that handles that. Would you be open to recommending each other to our respective users?” Include a specific co-promotion idea — a joint email, a mention in their newsletter, an integration.

Co-selling looks like: co-hosting a webinar on a topic relevant to your shared ICP, guest posts in each other’s newsletters, a simple integration that makes both products more useful, or just a mutual mention in onboarding emails. Even one partnership that reaches 500 warm leads with a 5% conversion rate is 25 customers.

Content That Drives Signups Early

Early-stage content that works is not thought leadership. It is specific, searchable, and solves a real problem your ICP has right now. The goal is not to build an audience. The goal is to create pages that rank for the exact phrases your customers type when they’re trying to solve the problem your product solves.

The highest-ROI content formats at early stage:

• Comparison posts: “[Your product] vs [alternative]” — high purchase intent, low competition if you’re new
• How-to guides: step-by-step workflows your ICP performs regularly that your product accelerates
• Problem-framing posts: naming a pain precisely that your ICP feels but hasn’t seen articulated well
• Case studies: real results from real customers, specific numbers, specific workflows

What doesn’t work early: broad “AI trends” posts, listicles about generic topics, content designed to build SEO domain authority over 6 months. You don’t have 6 months and you don’t have the domain authority to rank for competitive keywords.

Write one piece of genuinely useful content per week. Share it directly in communities where your ICP lives. Don’t blast it everywhere — share it when it’s relevant to a conversation. That single piece, shared in 3 targeted places, will typically outperform a generic content calendar of 5 posts.

The Launch Playbook: Product Hunt, HN, Reddit

These platforms matter but most founders sequence them wrong or over-invest in the wrong one.

Product Hunt: best for getting early feedback, building a small initial user base, and establishing a public presence. Not great for getting enterprise customers or anyone with a complex procurement process. Launch when you have at least 10 people who’ll upvote on day one — not because gaming upvotes is the goal, but because launches that start at 0 upvotes get buried in the algorithm before anyone sees them. Schedule your launch for Tuesday or Wednesday. Post between midnight and 2am Pacific. Have a clear GIF or demo video that shows the product working in 30 seconds.

Hacker News (Show HN): best for technical audiences and for getting noticed by sophisticated early adopters. The bar is higher — HN users are skeptical and the comments will find every weak point in your product and pitch. This is actually useful. Launch when you have a clear technical differentiator and you can explain it in one sentence. Be in the comments for the first 2 hours responding to everything. A top Show HN post can drive 500–2000 signups in 24 hours; an average one drives 20–50.

Reddit: the most underrated of the three for early customer acquisition, because subreddits are already segmented by ICP. Don’t launch with a post — join the community first (see community tactics above), then share when relevant. A well-timed post in the right subreddit with genuine context can drive more targeted signups than a Product Hunt launch.

Sequencing: community engagement and personal network first (weeks 1–4), then Show HN when the product is technically solid (week 5–8), then Product Hunt when you have polish and an existing user base to mobilize (week 8–12).

Converting Free Users to Paid

The moment founders dread most is asking for money. That reluctance kills more startups than bad products.

The conversion conversation is simple when you frame it correctly. You are not asking someone to do you a favor. You are asking them to pay for something that is already working for them. If it isn’t working for them, don’t ask — go fix that first.

The timing: after a user has had at least one “aha moment” — when they’ve used the product for its core use case and gotten a real result. Not after signup. Not after day 1. After they’ve seen it work.

The script: “You’ve been using [product] for [timeframe] — I wanted to check in. Has it been useful? What’s the one thing it’s helped you with?” Let them tell you the value. Then: “I’m glad that’s been useful. We’re moving to a paid model — it’s [price] per month for what you’re using now. Want me to set you up?”

Don’t ask for feedback after you’ve asked for money. Ask for the money. Silence after you make the ask is not a no. Wait it out. The founders who add “no pressure, just let me know what you think” immediately after giving a price systematically undermine their own close rates.

Realistic conversion rates from free to paid: with direct individual conversations, expect 20–40%. With self-serve upgrade flows and no personal follow-up, expect 2–8%. This is why doing the conversion manually for your first 50 customers is worth the time.

What to Do When You Have 10 Customers

Ten customers is not the goal — it is a research asset. You now have access to 10 people who trusted you enough to pay. Use that.

Talk to each of them in depth within their first two weeks of usage. Not a survey — a 30-minute conversation. Ask: what problem were you trying to solve when you found us? What almost stopped you from signing up? What do you use us for most? What would make you stop using us? What do you tell others when you describe what we do?

The answers to these conversations will rewrite your pitch, clarify your ICP, and surface the two or three features that matter most. They will also tell you the exact language your customers use to describe the problem you solve — which is the language you should use in your marketing copy, not the language you invented.

Do this before you try to grow. The gap between 10 customers and 100 customers is almost never about channel — it’s about clarity. Founders who talk to their first 10 customers and distill the real insight from those conversations grow faster than founders who are already optimizing a paid ads funnel on an unvalidated message.

At 10 customers you also have something you didn’t have before: proof. Three of those customers will say something specific and quotable about the value they got. Ask them if you can share it. That testimony — one sentence from a real person with a real name — is worth more in conversion than any feature you could ship in the same time.

The Summary That’s Actually a Checklist

Weeks 1–2: personal network outreach (50–80 messages), join 3–5 communities, do concierge delivery for the first 5 customers to close them.

Weeks 3–4: cold outreach to 100–150 targeted contacts, continue community engagement, convert free users to paid one-on-one.

Weeks 5–8: Show HN launch, one co-selling partnership, 2–3 pieces of high-intent content.

Weeks 9–12: Product Hunt launch, leverage early testimonials in all outreach, start tracking what’s repeatable.

The thing almost every founder gets wrong: they want to get to the scalable channels before they’ve done the unscalable work. The unscalable work is not a phase to skip. It is where you find out whether your product actually solves the problem, whether your ICP is who you think it is, and whether the pitch that converts is the pitch you started with. Almost always, it isn’t. And you can only learn that by doing the slow, personal, uncomfortable work of getting the first 100 customers one at a time.