How I Closed 5 Clients Using Claude Prompts

I closed five new clients in 11 weeks using Claude prompts I built myself. No sales team. No agency. Just me, a laptop, and a very specific system I put together after two years of trial and error running a solo real estate consultancy in Madeira. I’m going to show you exactly what I did.

Before I get into it: I’m not a sales person. I hate cold outreach. For most of 2023 and early 2024, my client acquisition was basically word of mouth and the occasional inquiry from my website. Comfortable, but fragile. Then I started treating Claude not as a writing tool, but as a client-closing system. That shift made all the difference.

The Problem Every Freelancer Knows but Rarely Admits

Most freelancers are good at the work. Terrible at getting the work. I was no exception. I’d spend 90 minutes drafting a proposal email, second-guessing every sentence. I’d follow up once, feel awkward, and drop it. I had a list of warm leads I hadn’t touched in months because I didn’t know what to say after the first “sounds interesting” that went cold.

The real problem wasn’t effort. It was that I had no repeatable system. Every prospect interaction started from scratch. Claude fixed that — not by writing generic copy, but by helping me build a process I could actually run alone, consistently, without burning out.

What I Actually Tried (and What Failed First)

My first instinct was to use Claude the way most people use AI: “Write me a proposal for a real estate consulting engagement.” The output was fine. Polished, even. But it had no personality, no specificity, and nothing that sounded like me. I sent three of those generic proposals. Zero responses.

Then I spent a weekend rebuilding my approach. Instead of asking Claude to write things for me, I started feeding it context and asking it to help me think. That was the actual turning point.

I also tested ChatGPT for the same workflows during weeks three through five. It’s capable, but I kept finding that Claude held longer context better — especially when I was working through a multi-email follow-up sequence and needed the tone to stay consistent across five separate messages. Claude did that more reliably. ChatGPT would drift.

The 5-Part Claude Prompt System That Actually Closed Deals

Here’s the exact framework I built. Each part serves a specific moment in the client journey, from first contact to signed agreement.

Prompt 1: The Diagnostic Email (First Contact)

Instead of leading with what I offer, I now lead with what the prospect probably needs. My prompt to Claude looks roughly like this:

“I’m a real estate consultant in Madeira, Portugal. I’ve just had an initial inquiry from [prospect description — e.g., a British retiree looking to relocate, budget around €400k, mentioned concerns about NHR tax status]. Write me a first response email that asks three clarifying questions to qualify the lead. The tone should be warm but professional. I want to sound like someone who has seen this situation before and knows exactly what questions matter. Do not use generic real estate phrases.”

The result: a personalized diagnostic email in about four minutes. Previously I spent 25–30 minutes on first-contact emails. I sent 14 of these over 11 weeks. Eleven got replies. That’s a 78% response rate, compared to roughly 40% before.

Prompt 2: The Pain-Point Proposal

After the diagnostic call, I’d drop my notes into Claude and ask it to identify the three main concerns the prospect had expressed — then write a proposal that addressed each one directly, before mentioning price. I gave Claude a 200-word brief and asked for a 400-word proposal maximum. Short proposals close faster. I know this from experience.

The prompt always included: “Do not start with an overview of my services. Start with their problem.” That instruction alone improved my proposal-to-meeting conversion from one in five to roughly one in three.

Prompt 3: The Objection-Handling Script

I built a Claude prompt that generates a short objection-handling guide for each specific prospect. I’d tell Claude the service, the price, and the objection raised (“they said they want to wait until autumn to visit in person”). Claude would give me three short, honest responses — not pushy, just practical. I’d pick the one that felt most like me and send it.

This saved me the worst kind of time: the 40 minutes of staring at a screen trying to figure out how to respond to a “maybe later” without sounding desperate.

Prompt 4: The Follow-Up Sequence

I asked Claude to write a four-email follow-up sequence for leads who went quiet after an initial call. The key instruction: “Each email should stand alone — no ‘as I mentioned in my last email’ language. Each one should offer something small of value: a market insight, a relevant question, a short observation about Madeira property trends.”

I used this sequence on six cold leads. Three responded. Two became paying clients. One of those was a €1,200 retainer that started from a follow-up email four in the sequence — an email I would never have sent on my own because I’d have given up after email two.

Prompt 5: The Agreement Summary Email

Once someone verbally agreed to work with me, I’d use Claude to write a plain-English summary of what we’d agreed — scope, timeline, price, next steps — before sending the formal contract. This “pre-contract clarity email” reduced back-and-forth by a noticeable amount. Two clients specifically mentioned it made them feel confident about moving forward.

My Real-World Experience Closing Clients in Madeira

Let me walk you through the specific case that made me confident this system worked.

In February 2026, I had a warm lead — a Dutch couple in their early 60s who’d found me through a LinkedIn post I’d written. They were interested in buying a holiday home in the Calheta area, budget around €550k. We had a 30-minute video call. It went well. Then silence for 17 days.

Old me would have sent one tentative follow-up, gotten no response, and mentally written them off. Instead, I opened Claude and ran my follow-up sequence prompt. I gave it context: what they’d said on the call, their main concerns (renovation costs, rental potential, whether property prices in western Madeira were still rising), and their approximate timeline.

Claude produced four follow-up emails. The first one referenced a specific question they’d asked about short-term rental regulations in Calheta and offered a brief, honest answer — about 80 words of practical information. No pitch. Just useful. I sent it on day 18 after our call. They replied within two hours.

Over the next six days, we exchanged five more emails. I used Claude for two of those — the objection-handling script when they raised concerns about the renovation risk on older properties, and the pre-contract clarity email once they said yes. Total time I spent on all AI-assisted communications for this client: approximately 55 minutes across three weeks. Estimated time the same work would have taken me without Claude: three to four hours, with worse results and higher anxiety throughout.

They signed. The engagement was worth €2,800 in consulting fees plus a referral they sent me six weeks later. That referral became client number five in my 11-week count.

I want to be clear about something: Claude didn’t close these clients. I did. The tool helped me show up consistently, sound like myself, and not disappear when things went quiet. That’s what it’s actually for.

Comparing the Results: Before and After Claude Prompts

Metric	Before Claude System	After Claude System (11 weeks)
First-contact email response rate	~40%	~78%
Proposal-to-meeting conversion	1 in 5	1 in 3
Cold leads re-engaged via follow-up	1–2 per quarter	3 in 11 weeks
Time per proposal email	25–35 minutes	4–8 minutes
New clients closed	1–2 per quarter (avg)	5 in 11 weeks
Tool cost (Claude Pro)	—	$20/month

What Claude Does Not Do Well — From Personal Testing

I want to be straight with you here, because this matters.

Claude is not good at knowing when to stop. If you don’t give it a strict word limit, it will over-explain everything. My first few proposals came out at 700–900 words when 350 words would have been more effective. I had to add “maximum 400 words, no bullet points, no headers” to every proposal prompt before the output became usable. That’s a real friction point if you’re in a hurry and forget to add those constraints.

It also cannot replace your knowledge of the local market. I had to correct several emails where Claude included generic European property market observations that were wrong for Madeira specifically — things like typical mortgage structures or rental yield assumptions that don’t apply here. You have to feed it accurate local context, or the output sounds plausible but wrong. For real estate in a specific market, that’s a serious issue. Every draft needs a review pass before it goes to a client.

And it will not track your leads for you. Claude is a prompt-and-response tool. If you want to know which follow-up email a prospect is on, you need a separate system — even just a spreadsheet. I use a simple Notion table for that.

What I’d Do Differently If I Started This Today

I’d build my prompt library in week one, not week four. I wasted almost a month writing prompts from scratch each time before I realized I should save the ones that worked. Now I have a Notion page with 12 saved prompts, labeled by use case. That alone saves me 20–30 minutes a week.

I’d also set a strict rule earlier: every Claude output gets one edit pass before sending. No exceptions. The ones I sent without reviewing — there were three — had small errors that I noticed after the fact. None were disasters, but that habit matters more as the stakes increase.

And I’d start the follow-up sequence sooner. I was waiting too long after calls — sometimes eight or nine days — before running my follow-up prompt. The leads that converted fastest were the ones I followed up within 48 to 72 hours. Claude makes that fast enough to actually do it.

Practical Summary: What You Can Take From This Today

You don’t need 12 prompts to start. You need two: a first-contact diagnostic email prompt and a follow-up sequence prompt. Build those first. Use them for the next 30 days. Track your response rate before and after. That’s the entire experiment.

Claude Pro costs $20 a month. If it helps you close one additional client — even a small one — it’s paid for itself many times over. For me, the ROI over 11 weeks wasn’t even close.

The thing that matters most is this: freelancers don’t lose clients because their work is bad. They lose them because they go quiet at the wrong moment, send a weak follow-up, or never follow up at all. Claude doesn’t make you a better consultant or a better freelancer. It makes you a more consistent one. In client acquisition, consistency wins.

If you want the exact prompt templates I use — the diagnostic email, the follow-up sequence, and the objection-handling script — sign up for the Solo AI Kit newsletter. I share one practical prompt pack per month, built from real workflows, not theory.