The Cold DM Strategy That Books 3 Client Calls a Week (AI-Written)

Cold outreach has a bad reputation, and honestly, most of it deserves it. The average cold DM reads like a mail merge: "Hey [NAME], I love your brand! I noticed you could use help with [THING]. I would love to hop on a quick call to discuss how I can 10x your results." Delete. Block. Report.

For a long time I avoided it entirely. I got clients through referrals and inbound, which worked until it did not. There were months where nobody referred anyone, nobody filled out my contact form, and the pipeline went dry. The feast-or-famine cycle is brutal, and eventually I accepted that the only way to break it was outbound.

So I tried a different approach. Instead of starting with what I wanted (a client call), I started with what they needed (a specific problem solved). And I used AI to do the research and draft the messages. Now I send 15 to 20 DMs per week, get replies on roughly 20 to 30 percent of them, and book several discovery calls most weeks. No templates that sound like templates. No copy-paste spam. Just relevant messages that start real conversations.

Here is how the system works.

The Research-First Approach

The reason most cold DMs fail is not the writing. It is the lack of research. When someone sends you a message that could have been sent to ten thousand other people, you can feel it. There is nothing specific. Nothing that proves they actually looked at your work or understood your situation. It is a form letter with your name pasted in, and it gets treated like one.

My approach flips the order. Before I write a single word, I spend five to ten minutes researching the prospect. I look at their last ten to fifteen social media posts. I read their website copy. I check their landing pages, their lead magnets, their email signup flow. I look at what they are selling and how they are selling it.

What I am looking for is a gap. Something specific that I can help with. Maybe their Instagram content is solid but their landing page has no clear call to action. Maybe they just launched a course but their email sequence is a single welcome message. Maybe they are posting consistently but getting almost no engagement because their hooks are weak.

The gap has to be real and specific. Not "your marketing could be better" but "I noticed your checkout page does not mention your money-back guarantee, and that is probably costing you conversions." That level of specificity is what separates a message someone reads from a message someone deletes.

This is where AI becomes useful. Researching fifteen to twenty prospects per week is time-consuming if you are doing it manually. So I built a prompt that takes the raw information I gather and turns it into a personalized observation and message draft.

Here is the prompt I use:

You are a freelance outreach strategist who writes cold DMs that
start conversations, not ones that pitch services. I am going to
give you information about a prospect. Your job is to identify one
specific, actionable problem they have and draft a short DM that
references it.

Prospect information:
- Name: [NAME]
- Business: [WHAT THEY DO IN 1-2 SENTENCES]
- Platform I am reaching out on: [INSTAGRAM / LINKEDIN / TWITTER]
- Their recent content topics: [LIST 3-5 RECENT POST TOPICS]
- Their website/landing page observations: [DESCRIBE WHAT YOU
  NOTICED — weak headlines, no email capture, unclear offer, etc.]
- Their audience size: [APPROXIMATE FOLLOWER COUNT]
- What I can help them with: [YOUR SERVICE IN 1 SENTENCE]

Write a DM that:
1. Opens with a specific observation about their content or
   business (NOT a generic compliment)
2. Identifies one concrete problem or missed opportunity
3. Offers a brief, specific suggestion they can act on immediately
   (give value before asking for anything)
4. Closes with a soft question — not "can we hop on a call?" but
   something that invites a reply
5. Total length: 3-5 sentences. No more.

Rules:
- Do NOT use words like "I'd love to" or "just wanted to reach out"
- Do NOT pitch your services directly
- Do NOT use exclamation marks
- Sound like a peer offering a genuine observation, not a
  salesperson working a list
- The goal is a reply, not a sale

I run this prompt for each prospect, then edit the output to make sure it sounds like me and the observation is accurate. The AI handles the structure and phrasing. I handle the truth. If the AI suggests a problem that does not actually exist, I cut it and find a real one. The message has to be honest or the whole approach falls apart.

A typical message looks something like this: "Hey Sarah, I saw your post about the course launch last week — the content breakdown was really well done. One thing I noticed on your sales page: the testimonials are all at the bottom below the fold, and most visitors will never scroll that far. Moving even one above your main CTA could make a noticeable difference. Have you tested different placements?"

That is it. No pitch. No "I do this for a living." Just a specific observation, a useful suggestion, and a question that is easy to answer. About 25 percent of the time, the person replies. And when they do, we are already talking about their business, not about whether they want to buy something from a stranger.

From Reply to Call

Getting a reply is step one. Turning it into a discovery call is step two, and it is where most people fumble. They get a response and immediately switch into sales mode. "Great question, I actually help businesses with exactly this. Would you be open to a 15-minute call?"

Too fast. You just earned a tiny bit of trust with a thoughtful message. Do not blow it by pivoting to a pitch in your second message.

My second message is always another piece of value. If they replied to my observation about their testimonial placement, I might share a quick example of a page that does it well, or mention a specific metric I have seen from similar changes. I keep the conversation going for two to three exchanges before I mention anything about working together.

When I do bring it up, it is casual and low-pressure. Something like: "I have a few more ideas for your launch page if you are interested. I could walk you through them on a quick call — no pitch, just ideas you can use whether we work together or not." That framing matters. "No pitch, just ideas" removes the risk of saying yes. Most people are not afraid of a call. They are afraid of being trapped in a sales presentation.

Here is the prompt I use to draft follow-up messages after I get a reply:

You are helping me continue a cold outreach conversation that has
gotten a positive reply. I need a follow-up message that deepens
the conversation and moves toward booking a call — without being
pushy.

Context:
- My original DM: [PASTE YOUR FIRST MESSAGE]
- Their reply: [PASTE THEIR RESPONSE]
- My service: [WHAT YOU DO]
- Additional value I can offer: [ONE MORE SPECIFIC OBSERVATION
  OR SUGGESTION RELATED TO THEIR BUSINESS]

Write a follow-up that:
1. Acknowledges what they said (show you read their reply)
2. Adds one more specific, useful insight
3. If this is the 2nd or 3rd exchange, suggest a brief call
   framed as "sharing ideas" not "pitching services"
4. Keep it 2-4 sentences

Rules:
- Match the tone of the conversation so far
- Do NOT use phrases like "I'd love to help" or "let me know
  if you're interested"
- If they asked a question, answer it directly before adding
  anything else

The Numbers That Actually Matter

Here are the numbers from running this system consistently, averaged across recent months.

DMs sent per week: 15 to 20. I do not send more because quality drops when I try to scale beyond that. Each message requires real research, and I would rather send 15 good messages than 50 generic ones.

Reply rate: 22 to 28 percent, depending on the week and the platform. LinkedIn tends to run higher, around 30 percent. Instagram is closer to 20 percent. Twitter is unpredictable.

Conversations that lead to a call: about 40 to 50 percent of replies turn into a booked call within a week. Not every reply is a warm lead. Some people just say "thanks for the tip" and that is the end of it. But a surprising number are genuinely interested in talking more.

Calls booked per week: varies, but usually in the 2 to 5 range. The consistency is the point. There are still slow weeks, but I no longer have months where nothing comes in at all.

Close rate from calls: about 30 percent. Not every call becomes a client, and that is fine. The calls that do not convert still build relationships. Two of my best clients this year came from people I had a call with months earlier who were not ready at the time but came back when they were.

The math works out to at least a few new clients per month from cold outreach alone, on top of whatever comes in from referrals and inbound. It has not eliminated the feast-or-famine cycle entirely, but it has made the famine months much rarer.

Why This Works When Templates Do Not

The entire strategy rests on one idea: treat prospects like people, not leads. Research them before you message them. Reference something specific so they know you actually looked. Offer something useful before you ask for anything. And when you do ask, make it low-risk.

AI does not replace the thinking. It accelerates the execution. I still choose the prospects. I still verify the observations. I still decide when to suggest a call and when to keep the conversation going. But AI handles the time-consuming parts — drafting personalized messages, structuring follow-ups, and making sure each message is concise and focused.

If you are a freelancer or service provider who has been avoiding outreach because it feels spammy, this approach might change your mind. It changed mine. Cold DMs do not have to be cold. They just have to be specific.

The outreach prompts I shared above are adapted from the cold outreach section in Spark, which includes prompt templates for prospecting, first-touch messages, follow-up sequences, and objection handling. The full pack has 105 prompts across every marketing channel — email, social media, landing pages, SEO, and more. It is $29, works with any AI tool, and comes with a 30-day guarantee.

I built it because writing good prompts from scratch every time is slow, and bad prompts produce bad outreach. The difference between a DM that gets ignored and one that books a call usually comes down to the first sentence. These prompts help you nail it.