On this page
"Hi, saw your comment on a post about AI. Clearly you're thinking about how to scale. We built a platform that handles the full loop: list building, enrichment and personalised copy in one workspace. How are you currently running your campaigns?"
You've received that message. You muted that message. And the fear of being the person who sent it is probably why you haven't sent anything at all.
Here's the thing: the problem with that DM isn't that it's cold. It's that it's fake. It claims to have seen a comment, then proves in the next breath that no human read anything. A good cold DM is the opposite. It's not a pitch. It's evidence you actually read their profile.
This page is the four sentences that turn a stranger into a reply, real DMs I sent that worked, one I received that didn't, the openers that get you muted, and the honest numbers.
Quick answer
A cold DM that gets replies is four sentences: a trigger that proves you know them (something not derivable from their name and job title), a line connecting it to you, a line of value, and one soft ask. Under fifteen seconds to read on a phone, no link, no pitch. Sent by a founder to someone already living the problem, it isn't spam; it's the fastest route to your first customers. MyPip drafts the trigger line for each morning's people, and the formula below works anywhere.
Why do most cold DMs get ignored?
Because they're about the sender. The tell is the first line: "I saw your post", "I noticed you're a director at", or your name spelled wrong. Openers like these prove nothing except that a tool found you, and people can now smell automation in the first six words. A message that's all about the sender earns the mute.
Here are the openers that get me muted instantly, and that I'd never send:
"I saw your post on X." "I saw you commented on Y." "I see you followed Z." "I noticed you're a director at [company]." And anything with the wrong name or the wrong company. The post-and-comment openers deserve a special mention because whole AI sales tools are built on them, and they don't survive contact with a real person: people don't remember their own comments and posts, so "I saw your comment" reads as fake even when it's true. You can reference something they posted later in a conversation. It is not an opener.
I once got two identical emails from two different senders at the same company, two minutes apart. Same pitch, same job title on both senders, asking to speak to a colleague who'd left eight months earlier, and the +tag in my email address showed exactly which bought list I'd come from. The irony is I'd have been genuinely curious about their product. The zero care killed the trust before I got past the subject line. For high-volume senders that maths still works: send ten thousand, get a few dozen replies, burn the domain, repeat. You can't play that game. You're a founder. Your name is the asset, and the small stuff isn't small. It's everything.
Isn't cold DMing just spam?
No. Spam is volume without research. A cold DM done properly is one person writing to another about something true. And here's the reframe that matters more than any formula: nobody knows you or your product exists yet. The only way they find out is if you tell them. To get people talking, you have to talk to people.
The fear of being "that guy" is just fear, and it's smaller than it looks. What actually happens if one person thinks you're spamming? Nothing. You move on to the next person. You're a door-to-door salesman and there are millions of doors, most of which have never heard a knock. The founders who get stuck treat every send as a verdict on their character. It's one message to one stranger, and the stranger forgets it faster than you do.
Two beliefs run through every DM that works. One: show you know them. Real research, not "you're a director of this so I'm writing to you". Two: keep it short and ask a genuinely curious question that sits near your product without being a pitch for it. That's the whole philosophy. It's also one motion inside the full first-customers guide, and it works a lot better once you've done the steps before it.
What are the four sentences of a cold DM that gets replies?
Four sentences: a trigger that proves you know them, a line of relevance that connects it to you, a line of value, and one soft ask. That's the whole structure. Keep it shorter than feels safe; the entire message should read in under fifteen seconds on a phone.
Everyone overthinks the trigger, because it's the opener and it's the sentence that matters most. The usual failure isn't effort, it's excess: too clever, too much information, trying to compress the whole pitch into the hello. Don't. Keep it simple. Show you know them, offer something, ask one soft question, stop typing.
The bar for the trigger is specific: say something about them that isn't derivable from their name and job title. Everybody leaves a digital thumbprint, and the people you want to reach, founders and decision-makers, want to be found. They give talks. They write articles and posts. They show up in company newsletters. That's your raw material. You've done it right when the reply is some version of "huh, I didn't know that was public" or "yeah, that's genuinely happening at our company right now". That reaction is the moment a stranger decides you're a person and not a sequence.
Real cold DMs, torn down
Two messages I actually sent, from my time at Plastic Bank, and both got replies:
"Hey Steven, if plastic were as valuable as diamonds, would we have such a significant problem with ocean plastic pollution? What would you say plastic is worth?"
"Hey Silvestre, the average person throws away 28kg of plastic every year. What do you do to reduce that number?"
Look at what they don't do. No pitch. No "I'm from Plastic Bank and we..." No claim to have seen a post. Each one is a genuinely curious question inside the other person's world, with one concrete detail (a valuation frame, a 28-kilogram number) that proves thought went in. The product enters the conversation later, after they've talked about the problem in their own words.
Now the one from my inbox, the message at the top of this page. It fails the same test three times over. The opener claims to have seen a comment, which is the tell, not the trigger. The middle is a feature list about their workspace, which is value to them, not to me. And the ask is a meeting-shaped question about my "campaigns" before I've said a word. Every sentence is about the sender. Four sentences, zero evidence a human read anything.
Same length. Opposite results. The difference isn't talent, it's whether the research happened before the typing.
Where should a founder send their first cold DMs?
Wherever your specific customer is active and half-expecting messages. The platform question is usually asked backwards: it's not LinkedIn versus X versus Reddit versus email as a ranking, it's where the person you're writing to actually spends their time. Show up where they are, not where a tactics thread told you to be.
The one warning I'll give is about email. Email is the hardest place to start: people already expect spam there, and behind the scenes you inherit deliverability problems, mailbox maintenance, paid data and setup costs before your first send. The numbers back it up. Across 7.5 million cold emails sent in 2025, Belkins measured an average reply rate of 0.45 percent. Email done well is the most lucrative channel there is, but it's a terrible place to learn. If you don't already know it, don't start there.
The social platforms carry a quiet bonus: every real conversation you have on LinkedIn or X or Instagram or Reddit also builds your profile there. People see your name before you ever write to them, and the second cold DM from a familiar face isn't really cold. None of this matters, though, if the person you're writing to is the wrong person. Before the platform question, work out exactly who you're writing to.
What reply rate should you actually expect?
Honest numbers: mass cold email averages 0.45 percent replies (Belkins, 7.5 million emails, 2025). Backlinko's study of 12 million messages found only 8.5 percent get any response, but personalised bodies lift replies by about a third, and a single follow-up adds around two-thirds more. Research is the whole game.
Read those numbers the right way. The averages are dragged down by the exact messages this page told you not to send; the muted openers are the 0.45 percent. You're not running ten thousand sends. You're writing a handful of researched messages a day to people who are active and close to the problem, and at that scale, with real triggers, your replies will look nothing like the mass benchmarks. The studies agree on the direction: the more evidence of reading, the more replies.
On following up: once, a few days later, with something new to say. Backlinko's data puts one follow-up at a 65.8 percent boost in replies, so it's worth sending, and the second follow-up is where pest territory begins. If two genuine messages get silence, the answer is no for now. Move on to the next door.
The expensive part of all this was never the typing, it's the reading: an hour of profile research for every five minutes of message. That's the exact split MyPip takes off your plate, with today's people found overnight and a first line drafted for each person as evidence they were actually read. You check it, adjust it, and send it yourself.