Search “cold DM templates” and you’ll find a thousand fill-in-the-blank scripts that all sound exactly the same, because everyone is using them. The person you’re messaging has seen the template before, probably three times this week. That’s the trap with cold DM templates: the moment one works, it spreads, and the moment it spreads, it stops working.
So this isn’t another script to copy. It’s the structure underneath the good ones, plus a few real examples, so you can write a cold DM that sounds like you and gets a reply because it could only have been sent to that one person.
Why cold DM templates stop working
A template’s whole promise is that you can send it to everyone. That promise is also its flaw. The more people who use it, the more your recipient pattern-matches it to spam, and the faster their thumb finds the delete button. Anything that feels mass-produced gets treated as mass-produced, no matter how clever the copy is.
The fix isn’t a better template. It’s writing a message that’s obviously one-to-one. That sounds like more work, and it is, but only by about ninety seconds per message, and those ninety seconds are the difference between a reply and silence.
The four parts that actually matter
Forget word-for-word scripts. Every cold DM that works has the same four beats, and you fill them with their specifics, not yours:
- A real reason you’re here. Reference the specific thing they said or did. This proves you’re a human who read their stuff, not a bot working a list.
- One sentence of relevance. Why you, why this matters to them. No life story.
- A low-stakes offer. Something useful they can take or leave, not a demand for their time.
- An easy out. “No worries if not” makes saying yes feel safe, which counterintuitively gets more yeses.
Three cold DMs that read like a human wrote them
To someone who posted asking for a tool:
Saw your post asking how people track replies across inboxes. I got so annoyed by this I built a little thing for it. Happy to show you if it’d help, or just tell you what worked for me before that. No worries either way.
To someone complaining about a competitor:
Totally feel you on the data quality thing, I bounced off the same tool for the same reason. Ended up building my own approach to it. Not trying to pitch you, but if you want I’ll share how I fixed it. Either way, good luck with the search.
To someone who just launched:
Congrats on the launch, the positioning is sharp. Quick thing I noticed that might help with early signups, want me to send it over? Fellow founder, no agenda.
None of these are templates. Each one only makes sense sent to that specific person. That’s the point.
What to never do
Don’t open with “Hope you’re doing well.” Don’t use their first name twice. Don’t paste a paragraph about your company’s mission. Don’t fake a compliment about a post you didn’t read, because that’s worse than no compliment at all. And don’t ask for a 30-minute call from a stranger. You haven’t earned thirty minutes; you’ve earned, at most, a reply.
If you want the deeper version of why this works, it’s the same idea behind cold outreach that doesn’t feel cold: warmth comes from being specific, not from sounding polished. The reason most founders can’t be specific at scale is that finding the right person and their real context eats all the time. That’s the part we built Unbound Compute to do, so the ninety seconds you spend writing are spent on the message, not the research.