Tim Ferriss Cold Email Template: Why It Still Gets Replies

The Tim Ferriss cold email template comes from his 2008 post on emailing busy people, and it is short by design: name something specific the person made, say who you are in one clause, prove you already did the work, ask one question they can answer in two sentences, and give them an easy out. The structure is public. The reason it works is not.

The template, written out

Subject: [the specific thing they made], one question

[First name],

I'm [who you are, one clause, not a résumé].

[One sentence proving you already used, read, or built on their work.
A detail that would be wrong if this email went to anyone else.]

One question: [something they can answer in two sentences].

If it's not worth your time, no reply needed. I'll keep digging.

[Your name]
[One line of proof: a link to the thing you're building]

That is the whole artifact. The versions ranking on page one rewrite the wording and keep the shape, which tells you the wording was never the asset. The original post, "5 Tips for E-mailing Busy People," is still the first result, and still the only one about the sender's behavior rather than the sender's phrasing.

The skeptic goes first

The skeptic: A template from 2008, for reaching people who now get four hundred emails a day. The channel changed underneath it. Why would the words survive?

What I see: They didn't. Every copy of it has rewritten the words. What survived is the constraint: you cannot fill in the proof line without having actually read, used, or built on something of theirs. The template is a filter wearing a script's clothes.

The skeptic: Or it's survivorship. Ferriss got replies because he was Ferriss, and because in 2008 a cold email was still a novelty.

What I see: The 2008 part is fair, and anyone quoting a reply rate from that era is quoting an artifact. But the method predates the reach: the post is him writing down what he did when nobody owed him anything. And the constraint is testable now, which is what matters more than provenance.

The skeptic: Testable how? "Include a specific detail" is what every outreach tool on the market already claims to do. There's a personalization field for it.

What I see: A personalization field drops a company name into a slot. That is not the same object. Here is the test I actually run: swap the recipient. Change the name and the address, send the identical email to someone else in the same role. If it still makes sense, it will not get a reply, merge tags or no merge tags. A real proof line breaks when you move it. Four seconds to check, and it predicts the outcome better than anything else in the draft.

The skeptic: Fine. Then it doesn't scale, and everything I need has to scale.

What I see: Correct. That's the finding, not the objection.

“

The template is a filter wearing a script's clothes.

”

What breaks when you send it at volume

The skeptic: So I put a model on the research. It reads their last five posts and writes the proof line for me.

What I see: I build with these models every day, and that gets you a line that is true and worthless. A model summarizing someone's public output produces the observation any reader would have: "loved your piece on X," plus a clause of evidence. Accurate, and unmistakably machine-made, because the thing being noticed is the obvious thing.

The skeptic: And a human notices what instead?

What I see: Something slightly off. A decision they reversed. A number in a changelog that doesn't match the marketing page. The feature they quietly stopped shipping. Noticing requires an opinion about what is interesting, and that is the part I still can't hand off. Not because models are weak, but because "interesting" is defined by the gap between what you expected and what you found, and the model has no expectations.

The skeptic: Then what is safe to automate?

What I see: Everything except the one line that has to be true. Finding the address, formatting, sending, the follow-up timer, and the inbox that catches the answer and routes it. That last piece is what AgentMail does with programmable inboxes, and it removes the tab-switching without touching the writing. It's the same split I build everything on: when I made ReadyToPost draft social posts and replies, the machine got the drafting and the schedule, and the judgment about who is on the other end stayed with a person.

The skeptic: And if I ignore all of this and send three hundred a day anyway?

What I see: Your reply rate drops, which you expected, and your domain follows it down, which you didn't. Mailbox providers score you on how people react, so a week of ignored sends costs you the inbox placement of the emails you actually cared about. That is why warming and monitoring a domain is unglamorous work that decides everything downstream. The Ferriss template has one useful property here: it is cheap to send and expensive to write, so it self-limits. Cap your daily volume at the number of research units you can honestly do. For me that's five to ten before the lines start sounding like a summary. Not three hundred.

The version I'd send tomorrow

Four rules, applied to the shape above.

Subject line: name the artifact, not the ask. "Quick question" is burned. Use the title of the thing: the repo, the episode, the pricing change. It reads like a reply rather than an opening.

The proof line does the work; give it a fact with an edge. Not "I loved the launch." Instead: the thing you tried, what happened, and where it stopped. A friction point you hit inside their product is worth more than three paragraphs of praise, because it can only come from use.

One question, answerable in two sentences. Not a call. Not fifteen minutes. Not "pick your brain." A busy person answers a question from their phone in the elevator; they do not open a calendar link from a stranger.

Leave the exit unlocked, and mean it. "No reply needed if this isn't useful" is the line most senders delete because it feels like surrendering leverage. It's the opposite: the only sentence proving you priced their time correctly.

Then, before you hit send, run the swap test. Replace the recipient with someone else in the same role and reread. If the email still works, don't send it: go back and find the one thing only that person could answer. That decision takes four seconds and it is the only one in the whole sequence that changes what comes back.

FAQ

What is the Tim Ferriss cold email template, exactly?

A short email with five parts: a specific subject, a one-clause introduction, one sentence proving you did the homework, one narrow question, and an explicit out. It comes from his 2008 post on emailing busy people. Every version circulating today keeps that shape and changes the wording.

Does it work for sales outreach, not just networking?

The structure does; the intent has to change. A sales version still needs the proof line to be unfakeable, which means researching the account rather than the industry. If your offer only makes sense sent to two hundred companies at once, this template will not rescue it; it will just make the failure more polite.

How long should the email be?

Short enough to read fully on a phone screen without scrolling: roughly 60 to 90 words. Length is a proxy for the real constraint: if you needed four paragraphs to explain the ask, the ask isn't specific enough yet.

Should you follow up if there's no reply?

Once, after about a week, adding new information rather than repeating the request: something you shipped, tried, or found since. A follow-up that says "just bumping this" asks for time twice while offering nothing the second time, which is the exact behavior the template was built to avoid.