From live product to paying customers: a working playbook

The product is live and nobody is paying for it. That is the normal state of things on launch day, and buying more attention will not change it. Attention cannot rescue an offer people do not understand.

Keep this document open next to your project. Work through it in order. Every section ends with something you can do today.

The sellable offer

Before anything else, write down three answers. If you cannot, you have nothing to sell yet, no matter how good the software is.

• Person. Who exactly has the problem? One kind of person you could describe to a friend, not a market segment.
• Pain. What is expensive for them right now? Expensive in money or in hours.
• Promise. What changes after they use your product? State it as a result, with a time attached.

I will carry one worked example through this whole document. Call the product ClientPack.

The promise names the deck, not the technology that produces it. An agency owner buys the thing they would otherwise have stayed up until midnight to write.

Do now: write your three answers in one sentence each. Read the promise out loud. If it needs a second sentence to explain, it is still too vague.

Find, help, show up, repeat

Four moves, in that order. Nothing here needs a budget.

1. Find people already looking. Search is the cheapest research you will ever run. People type their problem into a search box, post it in a forum, or ask it in a community. That text is the wording of your offer, handed to you.
2. Help by doing the result by hand. Do not send them a signup link. Produce the outcome yourself, manually, and give it to them. Slow and unscalable is the point at this stage: manual work is how you create proof that a result is possible.
3. Show up with the same proof. Take that one finished result and put it in front of the next person, then the one after. Do not write a new pitch for every place you show up.
4. Repeat until ten people stay. Not ten signups. Ten who keep using it.

The whole loop rests on one idea: search shows you what people want, manual work creates the evidence, and repetition brings the next buyer.

Do now: open the search box people actually use for your problem, whether that is Google, Reddit, or a professional community. Copy ten real phrasings of the problem into a file. Those are your words from now on.

Your first week

Turn the loop into one week of concrete activity. These are the targets I would set for a first week with zero customers.

• [ ] List 30 possible buyers, all of the same type. One type only.
• [ ] Start 10 conversations. Personal messages, not broadcasts.
• [ ] Produce 3 results by hand and give them away as work you did for that person.
• [ ] Ask 5 people a day to look at what you made.

The shape of that week: 30 people, 10 conversations, 3 people helped, 1 pays.

That is a practical weekly target and not a guarantee. It may take two weeks or four. The ratio is there so you know when to change the message rather than blame the week.

Research tells you what to say. Personal messages are what create a conversation this week rather than next quarter.

Do now: name the 30. If you cannot reach 30 without inventing them, your buyer definition is too narrow or you do not know where they gather. Fix that before sending anything.

How much of the product should be free

There is no universal answer, and copying whatever the last product you admired did is how people give away too much or too little. Decide by one question: how does a user reach a real result?

The rule underneath all four rows: give people enough time or usage to experience a real result, then charge for the next meaningful one.

A longer trial is a learning tool only when the product genuinely needs more days to prove itself. If you extend it because nobody is converting, you have bought yourself a delay and learned nothing.

Do now: pick the row that matches your product today and write down the date you will review the choice. The right answer changes as reliability improves.

Where payment begins

The point where you ask for money should sit immediately after the first real result, never before it.

For ClientPack it works like this:

1. The user uploads a real transcript. Free.
2. They get back a complete deck, watermarked. Free, and it is a genuinely finished document.
3. Paying starts when they want the watermark removed, or when they want a second client deck.
4. The price is $39 per month, or $390 per year.

By the time a card is entered the buyer has held the output in their hands. What they are paying for is the next job.

Watermarking matters more than it looks. It lets the free result be complete rather than crippled, and it still leaves a reason to pay.

Do now: write one sentence naming the exact moment your product asks for money, and one sentence naming the result the user has already received by that point. If the second sentence is empty, move your paywall later.

What counts as evidence

Your decisions improve as you learn from customers, and the method has to change as their number grows. Reach for the wrong method and you end up running split tests on a handful of users and acting on the result.

Use the strongest evidence you have and no stronger. Small numbers are clues, not certainty. One customer saying "I could not find where to upload" is a clue worth acting on immediately, but it is not a measurement.

Do now: find your row. If you are in the "watch" row, sit with the next signup while they use the product, live or recorded, and write down the first place they hesitate.

The customer loop

A signup only matters if the person gets value. Four states, in order:

1. Sign up. The account exists.
2. First result. They have viewed a complete deck.
3. Use it again. A second client project goes through the product.
4. Keep using it. They return whenever the problem comes back.

A habit does not have to be daily. For ClientPack the right rhythm is "whenever there is a new client to pitch", and someone who returns on that rhythm is retained even when the product sits idle between pitches.

The other branch teaches you more. Someone stops or cancels, so you ask why, you group the reasons, and you fix one problem: the one that appears most often.

Cancellations tend to fall into a small set of buckets:

• wrong buyer
• hard to start
• product problem
• weak value
• bad timing
• price

These are not equally actionable. "Wrong buyer" means fix your targeting, not your product. "Price" is often the polite answer covering one of the others, so keep asking.

Four things to track, and only these four at this stage:

• [ ] Signups who reach a result
• [ ] Time to first result
• [ ] People who return
• [ ] Why people leave

Do now: write the cancel question you will send. One question, sent by a person, within a day of the cancellation.

Pricing with three checks

Pricing is guesswork until you run three checks. Each one sets a different boundary.

1. What the result is worth to the buyer. That sets your maximum. For an agency owner, a client-ready deck produced today is worth some share of what the pitch itself is worth.
2. What the alternatives cost: other tools, doing the work manually, doing nothing at all. Alternatives set the buyer's expectations of what a price like yours should include.
3. What it costs you to deliver, counting payment fees, AI usage, storage, and the support you personally give. That sets your minimum, below which more customers make things worse.

ClientPack lands at $39 with those three checks applied.

Do now: write the three numbers down as a range: your delivery cost, the nearest alternative's cost, and your honest estimate of what the outcome is worth. Your price lives between the first and the third. If your delivery cost sits above what the outcome is worth to the buyer, change the offer before you touch the number.

What is left after you deliver

Revenue is not the money you keep. Here is the arithmetic on a single ClientPack subscription:

Inside that $6.10 product budget sit four things: the AI run, PDF rendering, file storage, and direct support. An estimated AI run costs $0.15 to $0.30, which tells you how much room the other three have.

Three caveats travel with that table. The illustration uses US Stripe standard card pricing plus Billing pay-as-you-go, so your own fees will differ. The $31.20 is not take-home profit, because everything outside the product budget still has to come out of it. And keep usage limits in place until you know what real customers actually cost you.

On the annual plan, $390 per year saves the customer $78, which is two months free. That is a real discount you are funding, so decide whether you want the cash up front more than you want the difference.

Do now: build this table for your own product with your own payment processor's published rates. Then find your single largest cost line and set a usage limit around it.

One customer result, four places

You do not need a new message for every channel. You need one customer result, described once, and then placed where your buyer already is.

Build the story from three parts:

• Who. The pain and the context. "An agency owner with a pitch due the same afternoon."
• Result. What changed, with a number and a time period attached.
• Proof. A quote from them, or the finished document itself.

Then reuse that same story in four places:

1. Search content. Write the page that answers the exact phrasing you collected in week one.
2. Communities. Post it where the problem is being discussed, as an answer rather than an announcement.
3. Referrals. Give the customer something easy to forward to someone with the same problem.
4. Personal messages. The same story, sent to one person at a time.

Do now: write your first customer story in three lines, one per part. If the "result" line has nothing concrete in it, go back to the customer and ask what changed for them before you use the story anywhere.

Small creators and communities

Someone else already talks to your buyer every week. Partnering with them is cheaper than building an audience from nothing, provided you keep the tests small.

Start with 3 to 5 small tests. The sequence for each:

• [ ] Pick a small creator or community whose audience contains the right buyer.
• [ ] Give them a demo account plus a real customer result they can show.
• [ ] Agree a small fee plus an extra reward tied to sales.
• [ ] Give them a unique link or code so the traffic is attributable.
• [ ] Measure who gets value, who pays, and who returns.

Right buyers beat more followers. A community of a few hundred people with your exact problem will outperform a large general audience, and it will cost you less to test.

If you pay a partner, say so publicly. Undisclosed paid promotion is a legal problem in most markets and a trust problem in all of them.

Do now: list five candidates and rank them by how precisely their audience matches your one buyer type. Approach the top one this week.

Before you spend on ads

Paid acquisition multiplies whatever you already have. If customers do not stick, it multiplies the leak. Run this check before the first campaign.

• [ ] Right person. You know exactly who buys.
• [ ] Clear promise. They understand the result before signing up.
• [ ] First value. They reach the real outcome, not just the dashboard.
• [ ] Come back. They use it again.
• [ ] Track source. You know where each customer came from.

If any answer is no, fix that before you write a single line of ad copy.

One more number before you launch: the maximum you are willing to spend to acquire one new paying customer. Decide it in advance, write it down, and treat it as a stop line rather than a target. Setting it afterwards means setting it to whatever you happened to spend.

Do now: answer the five questions honestly with the evidence method from your stage. Then write your maximum acquisition cost on the same page.

The 30-day plan

Four weeks, one job each.

Week 1: find

• [ ] List 30 possible buyers
• [ ] Have 10 conversations
• [ ] Join 2 communities where the problem gets discussed

Week 2: help

• [ ] Help 3 people get the result
• [ ] Measure how long it takes
• [ ] Ask one of them to pay

Week 3: learn

• [ ] See who comes back
• [ ] Interview every cancellation
• [ ] Fix one problem

Week 4: repeat

• [ ] Repeat the way that worked best
• [ ] Compare who stays
• [ ] Find the next 10

Do not just get 10 customers. Learn why they stay, because that reason is the only thing you can repeat.

Across all four weeks, keep the same four numbers in front of you: people who get a result, time to first result, people who return, and the reasons people leave. Leave the rest of the dashboard alone until those four are moving in the right direction.