How to Find Your First 10 Customers (Without an Audience)

To find your first 10 customers, pick one narrow group of buyers, find the people in it who are publicly showing the problem right now, write to each of them one at a time with a reason that comes from their own words, and count someone only when they pay. You do not need an audience or a big network. You need a list of people with a live problem and a first message that proves you read what they wrote.

This guide covers each step, including where to find those people on Reddit, X and LinkedIn and how to rank them. It assumes you have already checked that the problem is worth paying for. If you have not, start with how to know if people will pay for your SaaS and come back.

To keep things concrete, the examples follow one made-up product: a tool that turns support-ticket exports into a weekly report of why customers cancel, for small SaaS teams. Replace it with your own.

What counts as a customer, and why 10 is the number

A customer is someone who has paid you, or signed a commitment with a price on it. Write that rule at the top of your tracking sheet before you start. These do not count:

• Waitlist sign-ups
• Free trial users
• Friends who say they will "definitely use it"
• Pilots with no price attached

Ten is a useful target for a practical reason. One or two sales can be luck. By ten, you start to hear the same needs and objections repeated, and you have enough happy people to ask for introductions and references. Stripe's guide to early sales makes the same point: by the time you have ten happy customers, it stops looking like a fluke, and you start to see recurring themes in what they need.

It will not sustain a business, and it is not meant to. It is the point where you know who buys, why, and what words make them say yes.

Step 1: Pick one narrow buyer and one trigger

"Small businesses" is a category, not a customer. Your first buyers are easier to find, and easier to write to, when you can describe them in one sentence with a reason they need you now:

We help [specific person] who [recent trigger] get [specific result] without [their current workaround].

For the example product:

We help founders of small SaaS companies who just lost a few customers in a month find out why, without reading hundreds of support tickets by hand.

The trigger matters most. A person who has just lost customers, hired for a role, or complained in public has a reason to reply today. A person who merely matches your target description has no reason to reply at all.

Two quick checks before you move on:

• Can one person say yes? Founders and small-team owners can approve a purchase themselves. Buyers who need several sign-offs take months.
• Do they act in public? People who post, comment and reply online are the ones you can find with the methods below. Someone with no online footprint may be a good customer, but not one of your first ten.

Step 2: Find people who are already showing the problem

This is the step most guides skip. They say "find prospects" and move on. Here is what to look for and where.

The signals worth looking for

Not every mention of a problem means the person wants a fix. Learn to tell them apart:

Signal	What it tells you	Strength
Says they are looking for a tool, or asks for a recommendation	They want a solution now	3
Complains about a competitor's price or a missing feature	They are paying and unhappy	3
Describes a workaround ("I export it to a spreadsheet every Friday")	The problem costs them time	2
Posts a job ad for the manual version of the task	A company has set aside budget	2
Reacts to or reposts a competitor's post about the problem	They care about the topic, in public	1
Mentions the topic in passing	Interest at most	0

Keep the strength number. You will use it to rank your list in step 3.

Reddit: threads where people ask for help

People ask for tools and complain about the ones they use in subreddits for their job, hobby or industry. Search comments and posts for phrases that signal a live need, such as "is there a tool", "alternative to", "how do you handle" and "looking for". Then read the replies, because the people answering often name what they pay and what they switched from.

You can search by hand, or use the Reddit comment search endpoint to pull the results into a list. To watch a few communities for new questions, list a subreddit's newest posts once a day.

LinkedIn: posts and job ads

LinkedIn shows two kinds of signal. Post search finds people writing about the problem in their own words. Job search finds companies hiring someone to do the manual version of what your product automates. A listing that says "review cancellation reasons across tickets every week" is a company paying a salary for your problem, and the listing tells you which team owns it.

X: people asking, and people replying

X search accepts advanced operators, so you can look for "alternative to" a competitor, a complaint about their price, or a request such as "anyone know a tool that". Use the X search endpoint for the same queries at scale, and add filters such as a minimum number of likes to skip noise.

Your competitor's audience

People who engaged with a competitor in public have already told you they care about your category. You can collect three groups:

• People who reposted a competitor's launch or announcement (X retweeters)
• People who replied to it, including with complaints (X tweet replies)
• People who reacted to a competitor's LinkedIn post (LinkedIn post reactions)

The replies are the best of these. A person who writes "we stopped using this because of the price" under a competitor's post is a signal of strength 3 with the reason written out for you.

Adjacent tools: audiences of early adopters

Pick a tool your buyer already uses that does not compete with you. People who follow it, and especially those who reply to its posts, have already adopted a young product, so they are used to trying new things. Pull the account's followers with the X followers endpoint and keep the ones whose bio matches your buyer, such as "founder" or "CEO" with a company name. Stripe's guide recommends this approach, finding customers of lateral companies, because it removes the biggest objection to buying from a new company.

People who just switched

Search for phrases that show a recent purchase in your category, such as "we switched to", "we just moved to" and "we started using", followed by the type of tool. Someone who bought in the last few weeks has a budget and a buying process, and has shown they will try a newer vendor. They may not need you today, but they are good people to talk to about what they chose and why.

If you sell to marketers or developers

• Marketers and ecommerce brands: Companies running ads today have a marketing budget, and their ad copy shows what they are paying to solve. The Facebook Ad Library search tool shows who is advertising in your niche.
• Developer tools: The issue pages of popular open-source projects in your area are full of people asking for things the project does not do. Anyone writing "is there a way to…" there is a lead.

Run the search with a script

Reading results by hand works for a first pass. Once you repeat the same searches every day, a script saves hours. This one uses the Bridgly TypeScript SDK to search Reddit comments, keep those that sound like a live request, and drop anything older than two weeks:

// find-leads.mts — run with: BRIDGLY_API_KEY=bgly_... npx tsx find-leads.mts
import { Bridgly } from '@bridgly/sdk';

const client = new Bridgly({ apiKey: process.env.BRIDGLY_API_KEY ?? '' });

const queries = [
  'alternative to zendesk',
  'tool to see why customers cancel',
  'how do you track cancellation reasons',
];

// Phrases that show someone wants a fix now.
const askSignal =
  /is there a (tool|way)|looking for|any (tool|recommendation)|alternative to|how do you/i;
const DAY_MS = 86_400_000;
const MAX_DAYS_OLD = 14;

for (const query of queries) {
  const res = await client.reddit.searchComments({ query, sort: 'new' });
  if (res.type === 'error') {
    console.error(`${query}: ${res.status} ${res.message}`);
    continue;
  }
  const leads = res.data.comments
    .filter((c) => askSignal.test(c.body ?? ''))
    .map((c) => ({
      comment: c,
      daysOld:
        c.createdAt == null
          ? null
          : Math.floor((Date.now() - Date.parse(c.createdAt)) / DAY_MS),
    }))
    .filter(({ daysOld }) => daysOld != null && daysOld <= MAX_DAYS_OLD);

  console.log(`\n${query}: ${leads.length} recent requests`);
  for (const { comment, daysOld } of leads) {
    console.log(
      `- ${daysOld}d ago · ${comment.subredditPrefixed} · u/${comment.author}`,
    );
    console.log(`  ${comment.postTitle}`);
    console.log(`  ${comment.commentUrl}`);
  }
}

Reddit comment search costs 1 credit per request, so a few dozen searches cost a few dozen credits. X search costs 1 credit per result it returns. See pricing for what a credit costs. The same pattern works for the other endpoints in this step.

If you work in Claude or Cursor, you can skip the script. The Bridgly MCP server lets you ask for the same thing in plain language, for example "find people on Reddit in the last two weeks asking for a way to track why customers cancel, and list them with links". Read the results yourself before you use them.

What the data gives you is public profile and post information. It does not give you email addresses. You reach people where they posted: a reply under their thread, a LinkedIn message, or the contact link on their own site.

Step 3: Score your leads and build a 30-name sheet

A pile of links is not a pipeline. Put the best ones into a sheet with these columns:

Column	What to write
Name	The person and their company
Source link	The post, comment or listing where you found them
Their words	The exact phrase that shows the need
Signal	0 to 3 from the table in step 2
Days old	How long ago they posted
Contact route	Where you will write to them
Source type	Reddit, LinkedIn, X, competitor audience, adjacent tool, other
Status	Not contacted, messaged, replied, call booked, paid, not a fit

Freshness changes how you use the list. A request posted two days ago is a live need. The same request from six months ago has probably been solved, or forgotten. Use a simple rule:

Days old	What to do
0–3	Write today
4–14	Write this week
Over 14	Write only if the signal is a 3, and say you saw it later

Start with the best 30. Thirty people with a strong, fresh signal will teach you more than three hundred with a weak one.

Step 4: Write to people using their own words

The opening line of your first message should quote or paraphrase what they wrote. That one detail separates a message that gets a reply from one that gets ignored. Keep it short, ask one question, and do not pitch in the first message.

Which route to use depends on where you found them.

Option A: Answer first, pitch never

If you found the person in a public thread, the best first move is a genuinely useful reply with no link. Answer the question. If a free spreadsheet template, a checklist or a clear explanation solves it, give that. People notice who helps in public, and some will ask what you are building. Read each community's rules first, and do not drop links where they are not welcome.

Option B: A short direct message

For people you found on LinkedIn or X, or who replied to your public answer, send a short message. Three examples:

After a Reddit thread:

Hi [Name], I saw your comment in r/[subreddit] about tracking why customers cancel. I'm building a small tool for that and I'm trying to understand how people do it today. What do you use right now, and what's the most annoying part? Happy to share what I've learned either way.

After a job ad:

Hi [Name], I saw [Company] is hiring someone to review support tickets for churn reasons. I'm working on a tool that does that part automatically. Is that role replacing something you do by hand now, or adding a new process? I'd like to hear how it works for you.

After a complaint about a competitor:

Hi [Name], I saw your reply about [Competitor]'s pricing. We're building something for the same job that costs a lot less. What did you need from it that it didn't do?

Each one refers to something specific, asks a single question, and makes it easy to answer in a line. Do not use a template that looks the same for everyone. If you cannot find something real to refer to, that person does not belong on your list.

For benchmarks, one published guide puts cold reply rates for founders at roughly 2 to 10 percent, and warm outreach at 10 to 34 percent. Treat those as rough ranges, because they vary by audience, offer and message. A message that quotes the person's own words behaves more like a warm one than a cold one, but only your own numbers will tell you by how much.

Option C: A one-page note for your best five

For the five leads you most want, do some of the work for them. Read their last 20 or 30 public posts, pick the problem they mention most, and write a one-page note: what you noticed, why it is likely costing them, and one thing they could try this week. Send it with no ask beyond "does this match what you see?"

You can use the API to gather their recent posts quickly (for example, an X user's tweets), but write the note yourself. A person can tell when an automated summary was pasted in. Use this only for a handful of high-value leads. It takes real effort, and that effort is the reason it works.

Keep it respectful and legal

Send every message yourself and to people who have a reason to hear from you. Do not use the data to bulk-message strangers. Check each platform's rules on messaging, and if you email anyone, follow the law where you and they live. In the United States, the CAN-SPAM rules require an honest sender, a clear opt-out and a physical address, and other countries have stricter requirements.

Step 5: Follow up, talk, and ask for payment

Most replies come after the first message, not to it. People are busy, and silence usually means "not now", not "no". A simple schedule:

Day	Message
1	The first message
4	A short follow-up that adds one useful detail or example
9	Ask whether you have the right person or the timing is wrong
16	Close the loop and say you will not write again unless they want you to

Stop after that. Writing every day to someone who has not answered is how you get blocked.

On the call

Do not demo first. Ask about the last time the problem came up:

1. "Walk me through the last time you dealt with this."
2. "What do you use for it today, and what does it cost you in money or hours?"
3. "What have you tried that didn't work?"
4. "If you wanted to fix it, who would decide and pay?"

Then ask whether they can buy on their own. Stripe's guide notes that this is an uncomfortable question that you should get used to asking. If they cannot decide alone, treat it as "not right now" and move on to the next person.

Ask for the sale

If the product is cheap and one person can decide, ask for payment at the end of the call. For your first customers, a free trial often gives people an easy way to never start. Instead, offer founding-customer pricing, personal help getting set up, and a clear refund promise. A person who has paid has committed to using it, and you learn much more from them.

Step 6: Turn each of the first 10 into the next ones

Every customer is a source of the next ones. After they have used the product for a few weeks, ask:

• "Who else do you know who has this problem?"
• "What did you search for or read before you found me?"
• "Can I quote you, and can I name your company?"

The second question is the most useful. The words they used to look for a fix are the search terms you should use in step 2.

Then look at the source column in your sheet. After 30 days, count how many leads from each source type replied, booked a call and paid. The result usually surprises people. One source tends to produce most of the customers, and that is the one to repeat. Run the same search, with the same message, on a fresh batch before you try new channels.

Mistakes that stall the first 10

• Targeting everyone. A message that fits everyone gets a reply from no one.
• Counting free users. Free users are useful for feedback, but they do not tell you whether anyone will pay.
• Messaging in bulk. It gets you blocked and teaches you nothing.
• Pitching in the first message. Ask about their situation first.
• Giving up after one message. Most replies come to the second or third.
• Building more before anyone has paid. The next feature is rarely what stands between you and customer one.
• Using old signals. A request from six months ago is a story, not a lead.

A 30-day plan

Week	What to do	Number to hit
1	Write your one-sentence buyer. Run the searches in step 2. Build the sheet of the best 30.	30 leads with a signal of 2–3
2	Send the first messages, or public replies, to the freshest leads. Follow up on day 4.	20 messages sent, 5 calls booked
3	Run the calls. Ask for payment or a priced commitment at the end of each.	5 calls done, first payments
4	Follow up with everyone who hasn't answered. Ask each customer for introductions. Check the source column.	10 more leads from referrals; the best source identified

These numbers are targets, not promises. If you are well short after two weeks, look at the sheet before changing the product. Weak signals, old signals, and messages that don't quote the person are more common causes than a weak product.

Frequently asked questions

How long does it take to get your first 10 customers?

For a low-priced product that one person can buy, expect a few weeks of steady daily work. For a higher-priced B2B product with several people involved, expect a few months. The time depends mostly on how fast you can find fresh leads and how many calls you can hold each week.

How many conversations do you need to get 10 customers?

There is no fixed number, because it depends on the price and on how well the people match your target. Plan for dozens of real conversations, not a handful, and track your own conversion from conversation to payment after the first batch.

Should you give early users a free trial?

Usually not for your first customers. A trial lets people say yes without committing, and many never start. Founding-customer pricing, hands-on help getting set up, and a refund promise work better and tell you whether people will pay.

Can you find customers without an audience or a network?

Yes. Look for people who are already asking for a solution in public on Reddit, X and LinkedIn, and reach out with a message that refers to what they wrote. You do not need followers for that.

What if nobody replies?

First check your list. Are the signals strong and recent, and does each message quote something the person actually said? If yes, change one thing at a time: the audience, the opening line or the ask. If you have sent 30 well-targeted messages and have no replies at all, the problem may not be urgent enough for people to act on.

Sources

• Stripe Atlas, Your first 10 customers
• Prospeo, How to get your first 10 customers (cold and warm outreach reply-rate ranges)
• 100tasks, How to find your first 10 customers with no network (scoring leads and follow-up cadence)
• Federal Trade Commission, CAN-SPAM Act: a compliance guide for business