To find first customers for a new service, do not start with proof you do not have. Start with one sentence about who has the problem, give those people something small and useful before you ask for a meeting, look at real prospects before you send anything, and use the first replies to rewrite the offer. This article is a 30 day plan for the owner of a B2B service firm whose offer is defined, whose existing clients do not need it, and who has no case study yet.
The target is ten buyer conversations in 30 days. It is a target and not a promise. Its job is to give you a date on which you judge the offer with evidence.
We wrote the plan around the GTM Bud connector, because the campaign build happens in a chat with your own Claude or ChatGPT. Only the account setup is on gtmbud.com. We describe below exactly what GTM Bud returns at each step and what your AI writes, taken from how the connector is built.
Why is a new service harder to sell than your current one?
A new service is harder to sell because all three things that sell your current one are missing. You have no referrals for it, no client who can vouch for it, and no sentence that you have said a hundred times.
Take a 16 person accounting firm in Leeds. It does year-end accounts and payroll for owner-managed companies. The owner, Farah, has built a new service: a monthly cash forecast and board pack for companies with outside investors. Her existing clients have no investors. They do not need it.
Her situation on 1 November:
- The offer is defined. Scope, price and delivery are written down.
- The client list is no help. Nobody on it has the problem.
- There is no proof. One internal test on the firm’s own numbers is all she has.
- The deadline is real. She hired a part-time analyst from 1 December and must tell her business partner by then if the service has buyers.
Most owners in this position wait for a first client to appear and then build the case study. The plan below reverses the order: the conversations come first, and they shape the offer before the first client signs.
The 30 days at a glance
| Days | Job | You hand over |
|---|---|---|
| 1 to 3 | Audience sentence and criteria check | A draft campaign with criteria that passed |
| 4 to 6 | Prospect preview | Marks on real people, and a tighter sentence |
| 7 to 10 | Value offers that need no proof | One to three offers selected for the test |
| 11 to 13 | Messages, readiness check, launch | A live campaign that you confirmed |
| 14 to 27 | Reply review, twice a week | Notes on what buyers said, in their words |
| 28 to 30 | Decision | One page: keep, change or stop the service |
Days 1 to 3: What is the audience sentence for an offer nobody has bought?
The audience sentence names the role, the kind of company, the size and the place of the people who have the problem your new service solves. For a new offer, write it from the problem and not from your client list.
Farah’s sentence: “Finance directors and founders at companies with 20 to 150 employees in the north of England that have raised outside investment and have no CFO.”
She typed: “Create a campaign draft named Board pack launch for this audience, and show me the criteria check.”
Here is what happens. Her AI turns the sentence into criteria, one requirement in each, from one to seven of them. GTM Bud checks each criterion and saves the draft. The check looks for a requirement that cannot be read from a LinkedIn profile, a company page or public web content. It also flags a criterion that only repeats what GTM Bud sourcing does for every campaign.
In Farah’s case, one criterion was flagged: “has no CFO”. A profile shows what a person is, not what a company lacks. The check gave a message and a suggested rewrite. Farah made it a nice-to-have, which is a preference and does not reject a person. “Has raised outside investment” passed, because companies announce funding in public.
One more step belongs here. The seller profile of the draft needs the firm’s website and the owner’s personal LinkedIn profile. GTM Bud reads both. If your website does not mention the new service yet, add a page before day 7, because your AI will research the website when it writes the offers.
Days 4 to 6: What does a prospect preview tell you before launch?
A prospect preview is a search that returns real people who match your audience. It takes one to three minutes. It tells you if the sentence finds the buyers you imagined, before a single message exists.
Farah asked: “Start a prospect preview for Board pack launch and show me each person with title, company and headcount.”
- Good fit. Owen T., Finance Director at a 60 person software company in Manchester that announced a funding round last year. He is the person who builds the board pack by hand today.
- Not a fit. Sara L., founder of a 25 person company in Newcastle that is a subsidiary of a listed group. The group’s finance team does her reporting. Farah marked her and chose to exclude everyone like this.
Two rules for these three days:
- Read the full profile of each person. A preview is a short list, so this takes minutes. Ask your AI to open each one. The profile shows the result for each audience requirement, so you see why a person matched.
- Change the sentence, not your standards. If half of the preview is wrong, update the targeting of the same draft. A preview is one list for each run, and each campaign has a daily preview limit, so change the audience first and then run it again.
For a new service this step does a second job. If two or three rounds of changes to the sentence still return people you would not be glad to sell to, the service may have a market problem and not a marketing problem. It is cheaper to learn that on day 5 than on day 30.
Days 7 to 10: Which value offers work without a case study?
A value offer is a free and useful thing that the campaign offers to each lead. It works without a case study because the lead judges the thing itself, not your history. For a new service, the offer should be a small sample of the service.
This is the step where owners most often misread what the tool does, so here is the exact split.
What GTM Bud returns. Your AI asks for the value offer brief of the campaign. The brief holds the GTM Bud instructions for value offers, the seller profile, the audience, the offers the campaign already has, and the output format. It also holds two preferences. If you have not set them, your AI asks you to choose:
| Preference | Options | What it means for a new service |
|---|---|---|
| Format | Any, 1-on-1 call, Video, PDF or report | A report or a video shows the method. A call asks for trust first |
| Deliverable mode | Any, One suits all, Personalized | One resource for all leads, or one that you make for each lead who replies |
What your AI writes. GTM Bud does not write the offers in this flow. Your AI researches your company and the audience, follows the brief, and writes up to five offers. Each has a name, what it reveals to the lead, what it is based on, the deliverable, a plan to make it, and one line on why a lead would want it. Your AI shows them to you and asks which to test.
What GTM Bud does next. You select one to three offers. Your AI saves them, GTM Bud checks them, and if there is a problem it keeps nothing until the problem is fixed. GTM Bud tests the selected offers against each other.
Farah asked for a report format, personalized. Her AI wrote four offers. She selected two:
- The 13 week cash view. “Send your last three months of management accounts. We return a one page 13 week cash forecast with the two weeks where cash is tightest.”
- The board pack teardown. “Send last quarter’s board pack with the numbers removed. We return one page on what an investor director looks for and does not find.”
Neither one needs a client name. Both are a slice of the paid service, and both take the new analyst under two hours. That last point matters: with the Personalized mode, you make the deliverable for each lead after the lead replies, so the offer must be cheap for you to keep.
Three offers to avoid when you have no proof:
- “A free consultation.” It asks for an hour and shows nothing.
- An industry benchmark you have not built. If the data does not exist, do not offer it.
- A discount on the first engagement. It turns the first conversation into a price talk about a service nobody has seen.
Days 11 to 13: How do you launch without a mistake?
You launch in three moves: write the messages, pass the readiness check, and confirm a launch preview. Each move has a point where you can stop.
- Messages. Your AI gets the sequence brief, which holds the writing instructions that GTM Bud uses, and writes one message set for each selected offer. You read them. Remove every claim that assumes a track record. “We have built this for our own firm” is true. “Companies like yours rely on it” is not.
- Readiness. Ask: “Can Board pack launch go live?” GTM Bud returns a checklist. Each check is met or not, with what to do. A missing sending account is fixed on gtmbud.com, where you also connect your LinkedIn account and set the sending schedule.
- Launch. The first launch call returns a preview with the first message of each channel. The campaign goes live only after you agree in the chat. Launch needs an active plan or trial.
If your AI has other connectors, such as a notes app, it can keep a log of each decision there. That is your AI’s work with your tools. GTM Bud does not write that log or send a report on a schedule.
Days 14 to 27: How do the first replies change the offer?
The first replies are the market research you could not buy. Review them twice a week, for example Tuesday and Friday, and write down what each person said in their own words.
Ask: “Show me the replies in Board pack launch that need an answer, and open each conversation so I can see which offer the lead answered.”
Your AI reads each conversation, writes an answer from your reply playbook, and shows it for your review. GTM Bud instructs it to send only after you approve the exact text. A lead wrote to Farah: “We do the forecast in a spreadsheet. The pain is the pack, it takes me four days a quarter.” The AI wrote:
“Owen, four days a quarter is typical for a pack built by hand. The teardown looks at structure only, so you can remove every number before you send it. Shall I tell you which five pages we would need?”
Farah changed “is typical for” to “is a lot for”, because with a new service she has not seen enough packs to say what is typical. Then she approved the text.
Then sort what you heard:
| What the replies say | What it means | What you change |
|---|---|---|
| “We already do that part ourselves” | The offer shows the wrong slice | Lead with the other offer |
| “Not me, talk to our investor director” | The buyer is a different role | Change the audience on a duplicated campaign |
| Questions about scope and timing | The service is understood | Add the answers to the reply playbook |
| “What does it cost?” in most replies | The offer reads as a sales step | Make the deliverable more concrete |
| Not interested, with no reason | No signal yet | Wait. Do not rewrite for silence |
One limit to plan for. The connector changes value offers and messages on a draft only. For a live campaign, GTM Bud refuses the change in the chat and sends you to gtmbud.com, where you choose which leads get the new offer. So the clean way to test a rewritten offer from the chat is: duplicate the campaign, which gives you a new draft with the same targeting, value offers and messages, replace the offers and messages there, and launch the copy after its own preview.
By day 14, Farah’s replies pointed one way. Buyers kept the forecast for themselves and wanted help with the pack. She duplicated the campaign and rebuilt it around the teardown only.
Days 28 to 30: What do you decide?
On day 28, ask for the campaign results with the as-of time and read your notes. The results show replies by label: Interested, Warm, Not interested, and a count of replies with no label yet. Numbers can be up to about 15 minutes old, and the as-of time gives the exact moment.
Then write one page for your business partner with three lines:
- Who answered. The roles and company sizes of the people who replied Interested.
- What they wanted. In their words, not yours.
- What the service is now. The offer after the rewrite, and the price you will quote on the first call.
Ten conversations is the target. Four conversations that all say the same thing are also an answer. Zero after 30 days, with a preview full of the right people, is an answer about the offer.
For the wording of hard replies, see how to handle cold outreach replies. If your audience sentence feels vague, the ideal customer profile template helps you tighten it.
Frequently asked questions about finding first customers for a new service
How many people do I need to contact to get ten buyer conversations?
Nobody can give you that number for a new offer, and we will not invent one. It depends on the audience, the offer and the sender. Treat ten conversations in 30 days as a target that tells you when to stop and judge the offer, not as a forecast.
Should I offer the new service to existing clients first?
Yes, if they have the problem. Ask them first because it costs one email. This plan is for the case where they do not have it: the new service solves a problem that a different kind of company has, so your client list cannot tell you if it will sell.
Is a free first project a good way to get a case study?
A full free project is expensive proof, and it attracts companies that would not pay. A small free deliverable that takes you an hour and shows your method is cheaper and reaches more buyers. Charge for the first real engagement, even at a pilot scope.
Can I test two versions of the new service at once?
Yes, as two value offers in one campaign. You select one to three offers for the test, and GTM Bud tests the selected offers against each other. The campaign results in the chat do not split replies by offer, so you read which offer a lead answered in the conversation. Keep the audience the same for both, or you will not know if the offer or the audience made the difference.
What if nobody replies in the first two weeks?
Look at the audience before you blame the offer. Open the prospect preview again and read each profile as if you were about to call the person. If those people are right, the offer or the first message is the weak part, and you change that on a duplicated campaign.
Start with the sentence, not the case study
You find first customers for a new service by getting in front of the right people with something small and useful, and by letting their answers finish the offer. The case study comes from client one, and client one comes from conversations. To start, add the connector with the steps on GTM Bud for Claude, then read AI LinkedIn outreach for B2B service firm CEOs for the owner’s side of the build, or create an account at GTM Bud.