The standard advice is to build an audience and then sell to it. That works, it takes somewhere between one and two years, and it is usually given by people describing how their situation looks in hindsight rather than how it started. For the first ten customers there is a shorter route, and most people avoid it because it is uncomfortable in a way that posting is not.
The short route is a written list of specific people and a message sent to each one individually. That is the entire method. The difficulty is not in the sending.
- Ten customers is a number you can reach by hand. Reach is a problem for later.
- Building the list is the work. The message itself takes twenty minutes.
- With no track record you cannot promise a result, only offer to do the thing.
- What the first ten are worth to you is a case study, not the money.
- One kind of no is worth more than a yes, and people file all of them together.
Why audience first is the wrong order for this number
An audience is a distribution channel. Distribution is what you need when you want a thousand customers and cannot write a thousand messages. At ten, you can write thirty messages in a day, so the problem an audience solves has not arrived yet.
There is a less flattering reason the advice is popular. Publishing cannot fail on any given day. You post, some people see it, nothing measurable happens, and you post again tomorrow. Sending thirty individual messages produces roughly twenty-five silences inside a week, and those silences are countable and attributable. One of these activities protects you from information. The other hands it to you by Friday.
That does not make audience building useless. It compounds and it is worth starting. It is the wrong instrument for the first ten sales.
The list is the work
Open a document and write names. Not job titles, not segments, not marketing managers at mid-sized agencies. Names of people.
They come from narrower places than you expect. People you have worked with directly. People who complained about this exact problem in front of you at some point, which is the best source and the one everybody forgets, because the complaint felt like small talk at the time. People in a group or forum you already belong to. Former colleagues who have since moved somewhere the problem exists. People who bought something adjacent and told you why.
| Where the name came from | What you already know | What that lets you write |
|---|---|---|
| Worked with them | Their actual process, and the step that was painful | An opening line naming that step |
| They complained about this | Their own words about the problem | Their words quoted back, which is close to unanswerable |
| Same group or forum | The context you share, and why you are not a stranger | How you know them, in one clause, before anything else |
| Something they published | What they said they were dealing with | A reference to the thing they wrote, rather than a guess |
There is one test for whether a name belongs on the list yet. Can you describe that person’s situation in a sentence, without guessing? If not, they are not a prospect. They are research, and they go on a second list until you have done it.
Aim for thirty names to get ten customers. That ratio is not a benchmark and I am not going to present it as one. It is a number large enough that no single answer matters much, which is the only property that counts the first time you do this.
The discomfort of writing names down is the point. A list is countable. It turns a vague intention to find customers into a specific number of specific conversations that you have either had or avoided.
What you can honestly say with no proof
You cannot say it saves four hours a week. You have not measured that, and the first person who asks how you know will end the conversation.
You can say that you had this problem, that you built something for it, that it does one specific thing, and that you are looking for a few people to use it properly while you fix whatever breaks. All of that is verifiable and none of it is a claim about outcomes.
The offer at this stage is not a discount. It is access to you: someone who answers within the hour, takes the awkward edge case seriously, and changes the thing because they asked. That is worth more to an early customer than money off, and it is the one advantage you hold over every larger competitor for a while.
Being plain about the stage you are at also beats sounding established. The corporate voice is detectable, and it invites a comparison with companies that will win that comparison. Nobody expects a first version to be polished. They expect you to know that it is not.
What the message looks like
Four parts, in this order. How you know them or how you found them. One specific thing you believe about their situation. One sentence on what you built. A request small enough that agreeing costs nothing.
Written out, it reads roughly like this.
Hi Sam, we worked together on the Meridian rollout in 2024. I remember you rebuilding the weekly report by hand every Monday because the export dropped the column headings.
I got tired of the same thing and built a small tool that pulls the text out with the structure intact. It does that and nothing else.
Would you tell me whether it would have helped back then? Useful to me either way, and I am not asking you to buy anything.
The first sentence carries the message, because it contains something only someone who was there could write. Remove it and what is left is indistinguishable from every automated approach that person receives.
The common failure is asking for the sale immediately. A first message asking for money forces a decision the person has no basis to make. Asking whether you are right about their problem costs them thirty seconds, and anyone who answers has started a conversation.
What to ask for instead of money
Charge something. Free users are generous with encouragement and tell you very little about whether anybody wants this, and our piece on what to charge for the first thing you sell covers the floor a price has to clear.
Alongside the money, three things are worth more at this stage than the payment.
Permission to describe what happened, with specifics and their name attached. A vague testimonial is worth almost nothing. A sentence naming the company and the concrete change is what turns customer one into customer eleven.
One introduction. Not a general offer to refer you, which never materialises, but one named person they will actually email. Ask after something has gone well, and ask for exactly one.
A recorded conversation where they describe the problem in their own words. Their phrasing is better than yours and it becomes the copy on your page later. Ask permission, then use their language rather than translating it into yours.
How to read the answers
There are three kinds of no and they carry completely different information.
Not now. Timing. The budget is set, or they are mid-migration. Ask when to come back, write the date down, and come back on it. This is the closest a no gets to a yes.
Not me. The problem is real and they are the wrong person for it. The right response is a question: who deals with this? That converts a dead end into a warm introduction more often than people expect, because they are not turning you down, only redirecting you.
Not a problem. They do not recognise the thing you described. This is the expensive one and the most useful sentence you will hear all month. Four of these in a row means the product is the issue, not the outreach, and no amount of better messaging repairs it.
Most people record all three as rejection, then judge their progress by a number that mixes a scheduling conflict with proof that nobody has the problem.
Silence is a fourth category and it is not a no. It is a non-event, usually about their inbox rather than your idea. Follow up once, then stop, because a second follow-up converts an indifferent non-reader into someone with an opinion about you.
What we got wrong
Our first attempt described the tool. What it did, why it was built, what made it different. The reply rate was close to nothing, and the obvious conclusion was that the list was bad.
The list was fine. We rewrote the opening so the first sentence was about the recipient rather than about us, naming something specific to their work, and sent it to the same people. Replies came back from names that had ignored the first version completely.
The lesson is narrow and worth taking literally. The first sentence decides whether the rest is read, and it has to contain something you could not have written to anybody else. Everything after it is a rounding error.
What to do this week
Do not write the message yet. Open a document and get to thirty names, across several sittings if that is what it takes. Beside each one, write the sentence describing their situation. Any name where you cannot write that sentence goes on the research list.
Then send five. Not thirty, five, so you find out what is wrong with your opening line while it is still cheap to fix.
If what you are selling came out of a problem in your own work, the stage before this one is worth reading, and the capacity maths matters if you are doing this alongside a job, because outreach is the first thing dropped in a bad week.
Frequently asked questions
How many messages should I send before deciding it is not working?
Thirty, sent individually, before you conclude anything. Fewer than that and you are reading noise. What matters more than the count is whether the replies you do get are all the same kind of no, because thirty silences and thirty people telling you they do not have this problem are completely different results.
Email, LinkedIn, or something else?
Wherever that specific person actually reads things, which you usually know if they belong on your list at all. Email works when you have a real connection to name. A platform message works when the connection is the platform. The channel matters far less than whether the first sentence proves you know who you are writing to.
Is it worth running ads at this stage?
Not for the first ten. Ads buy reach, and reach is what you need at a thousand customers rather than at ten. At this stage you also cannot tell whether a poor result came from the ad, the page or the product, so you are paying for an answer you will not be able to read.
What if my list only comes to ten people?
Then send ten and treat it as research rather than sales. A short list usually means you are drawing only on people you know well. Widen it to people whose situation you can describe accurately, which is a much larger group than the people you feel comfortable messaging.
Should I say I built it myself?
Yes. At this stage it is the strongest thing you have. A named person who had the problem and built something is more credible than a vague company voice, and it explains why you are asking for their time rather than selling to them.
Someone said yes and then went quiet. What now?
Follow up once with something concrete they can respond to, such as a specific time or a direct question. If that goes nowhere, treat it as a no and move on. Enthusiastic silence is the most common outcome in early outreach, and chasing it costs more than the sale is worth.
How this was put together: this is the approach we used for our own tools, and the mistakes are ours, including the opening line problem above. We have quoted no reply rates or conversion figures, because they vary enormously by list quality and industry and a number invented for the sake of looking rigorous would be worse than none at all. Outreach rules differ in regulated industries and in some jurisdictions, which is worth checking before you send anything at volume.