You built it with Claude Code. Nobody is using it.

First, diagnose. Don’t build more.

“No users” has three different causes, and each needs a different fix:

Ask five people who fit your target customer one question: “How do you handle [the problem] today?” If they shrug, it is case one. If they describe a painful workaround, keep going.

Five questions for a diagnosis call

Use these with people who fit your target customer. Don’t show the product until the end.

1. How do you handle [the problem] today?
2. When did it last cause real trouble? What happened?
3. What have you tried to fix it? Why did you stop?
4. What does it cost you: in time, money or stress?
5. If it were solved tomorrow, what would change for you?

If answers to questions 2 and 4 are vague, the problem is not painful enough. But if people tell you a specific story with a specific cost, you are close. So ask for that story every time, and write it down word for word.

Stop shipping features for a week

Adding features feels productive. But with no users, there is nobody to tell you which feature matters. So the most useful thing you can do this week is ship nothing and talk to people.

Find the people already asking for help

Your future users are already describing the problem somewhere. Search Reddit, forums, Slack groups and LinkedIn for the problem in plain words, the way a customer would say it, not the way you would. Save 20 threads. Read how people describe the pain, what they tried, and what they would pay for.

Those phrases become your headline, your outreach and your landing page.

How to search for them

A few search patterns that work in Google:

• site:reddit.com “how do you” [problem]
• site:reddit.com “is there a tool” [problem]
• “[problem]” “any recommendations”
• “[current workaround]” “so annoying”

Save each useful thread in a simple sheet: link, the exact sentence that describes the pain, what they tried, and whether anyone answered well. After 20 threads, patterns appear. The same three complaints will repeat, and those are your headline and your first messages.

The 30-day plan

Week 1: one sentence and ten conversations. Write one sentence: “[Product] helps [specific person] [get specific outcome] without [specific pain].” Then have ten real conversations with people who fit it. Not demos. Conversations about their problem.

Week 2: one community, deep. Pick the single community where your people are most active. Answer questions there every day for a week. Mention your product only when it directly helps, and say you built it.

Week 3: a first paid offer. Offer a simple paid version to the people who replied. One price, one outcome. A real price tells you who has the problem badly enough.

Week 4: double down. Look at what produced replies and payments. Do more of that. Drop the rest.

Templates for weeks one to three

Week 1: asking for a conversation

“Hi [Name], I saw your post about [problem]. I’m trying to understand how people handle it today. Would you be open to a 15-minute call? I’m not selling anything yet.”

Week 2: a helpful community answer

“Here’s what has worked for others I’ve talked to: [step], [step], [step]. The common mistake is [mistake]. I built a small tool around step two; glad to share it if useful.”

Week 3: the first paid offer

“Thanks again for talking last week. I’ve turned what you described into [outcome]. I’m offering it to five people at $[price] per month, and I’ll set it up with you personally. Want one of the spots?”

Each template asks for one small step. Keep it that way.

Set up the minimum sales system

Before you drive more traffic, make sure the people who do show up don’t disappear:

1. A landing page with one form: one promise, one button.
2. A CRM or sheet: every signup or inquiry recorded, with the source.
3. A booking link: so interested people can talk to you today.
4. A follow-up reminder: nobody waits more than a day for a reply.

This takes an afternoon. It is the difference between 50 visitors who vanish and five conversations you can learn from.

A quick checklist

• Landing page with one promise and one button
• Form submissions arrive in a CRM or sheet automatically
• Each record stores the source (UTM tags or “how did you hear about us?”)
• A booking link people can use today
• A daily reminder of who is waiting for a reply
• A way to take payment, even a simple payment link

If any box is empty, fill it before you post anywhere new. Traffic sent into a broken path is traffic you pay for twice: once to get it, once to win it back.

What AI-built products that make money did next

The ones that reach revenue have little in common in their code and a lot in common after launch: a narrow buyer, a price from the start, founder-led conversations, and one channel worked deeply. We break down the pattern in the guide on vibe-coded apps that make money.

Common mistakes after launch

1. Treating launch day as the plan. A launch is one day. Distribution is every day.
2. Rebuilding instead of selling. A rewrite feels safer than a sales call. It rarely fixes zero users.
3. Talking about the tech. Customers care what it does for them, not that it was built with AI.
4. Copying a big company’s marketing. Brand campaigns need an audience you don’t have yet.
5. Pricing too low or not at all. It attracts the curious and hides who really needs it.
6. Changing everything at once. If you change the audience, the message and the price together, you learn nothing.
7. Quitting after two weeks. Most channels need a month of steady work before you can judge them.

Signs it’s working, and signs to change course

Rewrite your landing page in the customer’s words

Most AI-built landing pages describe the product the way the builder sees it: features, tech, a clever name. But a visitor decides in a few seconds whether the page is about their problem. So the first rewrite after your conversations is the headline, and it should use the words people used on the calls.

A simple structure that works for a first version:

1. Headline: the outcome, in the customer’s words.
2. Subheadline: who it is for and what it replaces.
3. One button: the single next step.
4. Three short points: the three problems people mentioned most.
5. Proof: one real quote or result, as soon as you have one.
6. A short FAQ: price, setup time, what happens to their data.

Illustrative before and after:

Before: “AI-powered workflow automation for modern teams.” After: “Send every quote the same day, without retyping it from your notes.”

The second headline is less impressive and much clearer. Clear wins.

Your first week, day by day

One week, no new features, and you will know more about your market than a month of building taught you.

Should you mention that it is AI-built?

Customers rarely care how the code was written. They care whether it works, whether their data is safe, and whether you will still be there next month. But some buyers have heard stories about insecure AI-built apps. So don’t lead with “built with AI”, and don’t hide it either. Lead with the result, and be ready to answer three questions plainly:

• How is my data stored, and who can see it?
• What happens if something breaks?
• Can I export my data and leave?

If you can answer those well, the build method stops mattering.

Before you charge: a trust checklist

Charging money changes what customers expect. Check these before you send the first invoice:

• Sign-in is secure, and passwords are never stored in plain text.
• Payments run through an established provider, not your own code.
• Customer data is backed up, and you have tested a restore.
• You have a privacy policy and terms that match what the product actually does.
• You know which third-party services see customer data.
• Someone other than you has tried to break the sign-in and payment flow.

If you are unsure about any of these, ask a developer to review those parts. It is cheaper than losing your first customers’ trust.

A worked scenario: from zero users to first payment

Illustrative example. A founder used Claude Code to build a tool that turns voice notes into client-ready meeting summaries. He launched on a few directories and a launch site. He got 120 visitors and four signups, and nobody came back.

Diagnosis. He called the four people who signed up. Two were other builders, curious about the tech. One was a consultant who liked the idea but didn’t use it. One was a real estate agent who said: “I record client calls in my car, and I never get the notes into my CRM.”

The change. That one sentence was more useful than the launch. He narrowed the product to real estate agents, changed the headline to “Your client call notes, in your CRM before you get home”, and stopped adding features.

Finding people. He found agents asking about note-taking and CRMs in real estate forums and groups. He answered questions for a week and sent 25 short messages to agents.

First payment. Five agents agreed to a call. Two tried a paid month. One kept paying.

One paying customer is small. But it proved the problem, the buyer and the message. So the next month was about doing more of the same, not building more.

Writing the one-sentence pitch

The one sentence from week one does more work than any feature. It goes into your outreach, your headline and every community answer. But most first drafts are too wide (“helps businesses grow”) or too technical (“an AI agent with RAG over your docs”). So test it against three questions:

1. Could a stranger tell if it is for them? If not, name the person more precisely.
2. Is the outcome something they already want? Use a result they would say out loud, not a feature.
3. Does it name the pain it removes? “Without retyping”, “without chasing”, “without a developer”.

Illustrative examples:

Read your sentence to someone who fits the description. If they say “that’s me”, keep it. If they ask “so what does it do?”, rewrite it.

Choosing the one community to go deep in

Week two asks you to pick one community and stay there. Picking well saves a month. Look for these signs:

• People ask for help, not just share news. Questions are your chance to be useful.
• Your problem comes up without you raising it. Search the community for it first.
• Members are buyers, not only builders. A builder community is great for feedback and poor for customers.
• The rules allow you to mention your product when it is relevant. Read them before you post.
• Posts get replies within a day. A quiet community won’t teach you much.

Once you pick, behave like a member, not a marketer. Answer fully, share what you learn, and thank people who help you. After two weeks, people will recognize your name. That is when your product starts to get a fair look.

How to read the first numbers

With only a few visitors, numbers can mislead. Use them as signals, not verdicts:

• Visitors but no signups: the headline or page doesn’t match what people came for.
• Signups but no first use: onboarding is too long or unclear.
• First use but no return: the result wasn’t strong enough, or it isn’t a frequent problem.
• Return but no payment: the offer, the price or the trust signals need work.

Each stage points to a different fix. Fix the earliest broken stage first, because every stage after it depends on it.

What to do this week

1. Call or message every person who signed up so far.
2. Decide which of the three cases you are in.
3. Write one sentence about who it is for and what it fixes.
4. Find 10 threads where those people describe the problem.
5. Rewrite your headline using their words.

Questions and answers