How It Works Results About What You Get FAQ Guarantee Blog Get Your Cold Email Playbook
The Playbook

The Cold Email
Playbook

The basics you have to get right, then the advanced work that separates a 1 percent reply rate from a 4 percent one. Written for anyone sending cold email, not just our clients.

Almost all cold email advice is about what to write. Writing is the smallest part of it.

Who you email decides whether anyone should want to reply. Your setup decides whether the email reaches an inbox at all. Your sequence decides how many chances you get. What you do with a reply decides whether interest becomes a conversation. Writing comes fourth.

The first half below is the basics. Get these wrong and nothing else matters. The second half is the advanced work, and it is where most of the difference between a campaign that books meetings and one that does nothing actually lives.

Part One: The Basics
Section 01

Who you email

This is the biggest lever in cold email by a wide margin, and it is the one most people spend the least time on. A brilliant email to the wrong person does nothing. An average email to exactly the right person works.

Be narrower than feels comfortable

Pick a group specific enough that you could write one sentence that only makes sense to them. If your list is "B2B companies", you cannot write that sentence. If it is "agency owners with 5 to 20 staff who sell retainers", you can.

Filter on 3 things: job title, location, and company size. Everything else is decoration at the start.

Email the person who can say yes

In a small company that is usually the owner. In a larger one it is whoever owns the outcome you are talking about. Emailing someone who has to forward your message to their boss adds a step where most campaigns die.

Split your list by company size

Run 5 person companies and 50 person companies as separate campaigns, not one blended list. They reply at different rates, they need different messages, and blending them hides which one is working.

Know when the list is the problem

Two numbers tell you. Fewer than 1 in 100 people replying, or fewer than 3 in 10 of those replies being positive. Measure once you have sent about 5,000 emails, because below that you are reading random noise.

When either one breaks, change who you are targeting before you touch a single word. It is almost never the writing.

Section 02

Getting delivered

None of the rest matters if the email lands in spam. This is the part beginners skip and then spend months confused about why nobody replies.

Never send cold email from your main domain

Not once, not to test. Your main domain carries your real email: clients, customers, invoices. Cold email damages the reputation of whatever domain sends it. That is unavoidable, which is exactly why you send from domains you can afford to lose.

Buy separate domains just for sending

Get lookalike versions of your real domain. If you are acme.com, buy getacme.com or acmehq.com.

Use .com. The cheaper endings like .info are the ones spammers use most, so filters already treat them with suspicion. You save a few dollars a year and inherit somebody else's bad reputation.

Treat them as disposable. When one stops performing you replace it, you do not try to rescue it.

Set up the 4 domain records

Every domain has a settings panel that tells the rest of the internet how to handle its email. Four entries matter.

  • SPF lists which servers are allowed to send email as you.
  • DKIM adds a hidden signature to each email so the receiver can confirm nobody tampered with it.
  • DMARC tells receiving providers what to do if either of the first two fails.
  • MX points incoming mail at your inbox, which is how you get replies.

Do not type these in by hand. Point your domain at your email provider and it writes the first 3 for you. Hand editing is where nearly everyone breaks something.

Warm up before you send anything

A brand new domain sending 50 cold emails on day one looks exactly like a spammer, because that is what spammers do.

  • 14 days of warmup with zero cold emails going out. Not fewer. Zero.
  • Do not connect a new account to a live campaign until day 15.
  • Then build up: 10 a day, then 15, then settle around 20.

20 emails per account per day

That is the ceiling. Want more volume, add more accounts, not more emails per account. Pushing a single account harder is how you lose the domain.

Send between 8am and 5pm on weekdays, in one time zone.

Turn off open and link tracking

Tracking works by rewriting your links to point at your email tool's own address first. That address is shared with every other customer of the tool, including the bad ones. When it gets blacklisted, every email you send is carrying a blacklisted link.

Turn both off. You lose open rate data, which was never reliable anyway, and you remove an entire category of risk.

Section 03

What to write

65 words, maximum

One sentence per paragraph, with a blank line between each one.

About 8 in 10 cold emails are read on a phone. Three sentences stacked together look like a wall of text on a phone screen, and the thumb scrolls past before the third one registers. Same words, completely different result.

The best performing email we have ever sent was 38 words. Shorter keeps winning further down than most people are willing to test.

Subject lines: 2 to 4 words, all lowercase

No full stop at the end. It reads like a note from a person instead of an announcement from a company, and it is the single biggest factor in whether the email gets opened at all.

Avoid anything that sounds like a prize draw. "You've been selected" and "we want you" both read as a scam before the email is even open. Avoid the word "open" as well, as in "open to a chat", because it sits on enough spam filter lists to be not worth it.

Ask for something small

The mistake almost everyone makes is asking for time in the first email. A meeting is expensive. Nobody gives a stranger 30 minutes.

Ask to send them something instead. Information costs nothing to accept, and accepting it is what starts the conversation. Once they have replied once, the meeting is an easy second step.

Give before you ask

The strongest cold emails describe something the reader receives, not something you could do for them. "We could put a plan together for you" and "here is the plan" describe the same thing and land in completely opposite places. One sounds like a proposal. The other sounds like a gift.

Put the ask at the end of the sentence about what they get, never on its own line. On its own line it reads like a demand. Tucked onto the end of the good part, it reads like a detail.

Two examples

Version 1

Subject: quick idea

{First}

I run a show for agency owners and consultants, one person per episode on how they actually grew the business.

What you have built at {Company} is the kind of story the audience asks for.

The episode is about you, your story, and the way you do things differently.

Can I send over some more info?

Jordan

Version 2

Subject: your work

{First}

I host a show for {audience} and I am booking guests for the next block of episodes.

You are one of the few people I found who {one thing}.

45 minutes, no script and no slides, and you keep the recording.

Can I send over some more info?

Jordan

The words in curly brackets get swapped in automatically for each person, using what you already know about them from building the list.

The formatting rules, and why they exist

  • No dashes used as punctuation. The most reliable sign of AI written text there is. Commas and full stops do the same job.
  • No exclamation marks. Enthusiasm from a stranger reads as a sales pitch, and filters score it that way too.
  • Write out "percent" and avoid currency symbols. Both are common filter triggers, and a money symbol in a first email signals a pitch before the sentence is read.
  • Avoid hyphenated words. Write "90 day" rather than joining them. Stacked hyphenated words read as machine written.
  • Use numerals. "3" rather than "three". It reads faster and it is how people actually type.
  • Just their first name, or no greeting. "Hi {First Name}" is the pattern everyone has learned to spot instantly.
  • Sign off with your first name only. No job title, no company, no closing line.
Section 04

The sequence

3 emails, on day 0, day 3, and day 6. Then stop. The discipline is in what each one is allowed to do.

The second email is not "did you see this?"

"Just checking whether this was of interest" is the most wasted email in cold outreach. It asks someone to make the exact same decision again, with nothing new to help them make it.

Give them something they did not have the first time. Proof is usually the strongest option: name companies who already said yes, written into the email rather than behind a link.

No link, on purpose. A link is more work, not less. It is a click, a new tab, and a decision about whether to bother. It trades the reply you want for a website visit you cannot use.

The third email is the goodbye

It does not repeat what the second one offered, so the two never overlap. It closes things out, applies no pressure, and asks one question that is easy to answer either way.

Email 3, the goodbye

Subject: closing the loop

{First}

I have not heard back so I will assume the timing is off.

Closing the loop on my end.

Worth circling back later in the year, or should I leave it?

Jordan

Longer sequences do not work better

Six and seven email sequences are common and they mostly generate complaints. By email 5 you are no longer following up, you are pestering, and every extra email raises the odds of a spam report that damages the domain you send everything else from.

3 well spaced emails get you nearly all the replies a longer sequence would, without the damage.

Stop for the right reasons

  • Stop when they reply. Any reply. Your tool should do this automatically, and you should check that it actually is.
  • Stop on an automatic out of office reply, and pick it back up when they are back.
  • Stop permanently on a clear no, and never load them into another campaign.

Following up after a positive reply

If someone shows interest and then goes quiet, a short follow up sequence is worth running: 5 emails spread over about 2 weeks, roughly every third working day.

Two rules keep it from turning into a problem. Stop the moment they book or clearly decline. And never add a sixth email. A sequence that keeps growing is a sequence that stopped respecting the reader.

That is the foundation

Everything above is enough to run a campaign that works. What follows is what separates a campaign that works from one that works well.

Book A Call
Part Two: Advanced
Section 05

List quality at scale

At a few hundred emails a month, list quality barely shows up. At tens of thousands, it becomes the thing that decides whether you can send at all.

Verify with 2 services, not one

Email verification tools tell you whether an address is real before you send to it. They disagree with each other more than you would expect, so run every address through 2 of them and keep only what both call safe.

Expect to lose roughly 2 out of every 3 addresses you bought. That feels like waste until you price the alternative, which is enough bounces that email providers stop delivering your mail at all.

The catch all trap

Some companies accept email sent to any address at their domain, real or not. A made up address gets accepted exactly like the CEO's. Verification tools cannot see through this, so they mark the address unknown. Some tools will then guess.

Whether that guess is safe depends almost entirely on company size, and the gap is bigger than most people expect.

What we sentEmailsBounced
Guessed addresses, companies of 1 to 20 people7,1962.5%
Guessed addresses, companies of 50+ people85217.4%
Fully verified addresses, companies of 50+ people1,4493.0%

Large companies are perfectly safe to email, as long as the addresses verify normally. The problem is only the guessing. At larger companies roughly 1 in 6 guessed addresses turned out to be dead, which is enough to get a campaign shut down.

So at bigger companies, drop the unknown addresses instead of guessing. You lose names. You keep your ability to send.

Watch what your filters delete

Every list gets filtered by job title, and every title filter quietly removes people you wanted.

The classic example: filtering out anyone whose title contains "coach" also removes "Founder And Coach" and "CEO & Executive Coach". Those are owners, which is usually exactly who you wanted. If a title says someone owns the business, that should beat every other word in it.

The other one worth knowing: an employee count of 0 nearly always means the data is missing, not that nobody works there. Treat it as unknown rather than filtering those companies out, or you will lose real businesses with real websites.

Once a month, look at a sample of what your filters removed rather than only what they kept. It is the only way to catch this.

Do things in the right order

Verify the addresses, then score the leads, then research them.

Any other order means paying for work you throw away. Research first and you are paying to look into someone whose email was never going to arrive.

Section 06

Deliverability at scale

How many accounts per domain

2 to 3 email accounts per domain is the safe standard. 5 is the aggressive end.

This is a real tradeoff, not a settled question. More accounts means more sending from each domain you paid for. It also means that when the domain gets flagged, you lose all of them at once and with a louder signal attached. We sit near the safe end. Anyone who tells you the answer is obvious is selling something.

Do not buy the cheapest accounts you can find

There is a tier of resellers offering bulk email accounts at a fraction of the normal cost. Those accounts get suspended at 15 to 30 percent a month. Proper accounts get suspended at 3 to 7 percent.

A suspended account mid campaign costs you more than the saving, every time.

Test every week and have a hard floor

Free deliverability checkers let you send a test email and see how many of the major providers put it in the inbox instead of spam. Run one on every sending domain, weekly.

If less than 60 percent is reaching inboxes, stop that domain and move to a new one. Do not wait for bounce reports. By the time bounces appear, your email has been going to spam for days.

Healthy looks like about 3 emails in 100 bouncing.

The tracking risk in numbers

This is why turning tracking off matters more than it sounds. When we audited our own sending accounts, 123 of 258 were pointed at a tracking address that had been blacklisted.

It affected nothing, because we had tracking switched off and no links in the emails, so not one message pointed at it. Had tracking been on, every one of those accounts would have been sending blacklisted links for weeks.

Let new prospects take priority over follow ups

Once you are sending at volume you will run out of daily capacity. Something has to give.

Send to new people first and let the follow ups wait. A new conversation is worth more than another nudge on someone who has already ignored you twice.

Section 07

Testing properly

Who it comes from beats what it says

This is the biggest finding we have, and almost nobody tests it.

We ran identical emails, to the same lists, over the same weeks, under 2 different senders. One was the founder. The other was a producer writing on the founder's behalf.

The founder version got 3.4 replies per 100 people. The producer version got 1.2. Every single founder version beat every producer version, by between 3.5 and nearly 6 times.

That is a bigger difference than any wording change we have ever tested. If you are choosing between rewriting the email and changing who it comes from, change who it comes from. Send as the founder.

The variation trap

Most sending tools let you write a few options for each line and have the software pick one, so every recipient gets a slightly different email. Everyone assumes those options multiply. They do not.

The software makes its picks from the same starting point for every line in a given email. Write 4 lines with 2 options each and you do not get 16 different emails. You get 2. Everyone receives either all the first options or all the second ones.

To get real variety, use a different number of options on each line. 3 options on one line, 4 on the next, 5 on the next gives you 60.

Kill things earlier than feels right

Kill a version as soon as it is losing on every measure, even before the numbers are technically conclusive. Waiting for statistical certainty costs weeks of sending to confirm what the data already showed you.

Two we killed:

  • A version sent to 573 people that got zero positive replies. We would normally expect about 6. Dead.
  • A version that mentioned the price. Across about 14,300 emails it got fewer replies (4.1 per 100 against 4.7) and fewer positive ones (1.3 against 1.5). Neither gap is conclusive alone. Both point the same way, so we killed it. Price does not belong in a cold email.

Be ruthless with subject lines

Our best subject line got a reply from about 4 people in every 100. We deleted 2 that were getting nearly 4 per 100, at a time when our best were running above 6.

Cutting a subject line that technically works feels wasteful, until you notice it is costing you 40 percent of the replies your best one would have brought in. Judge everything against your best, not against zero.

Test one thing at a time, and give it 5,000 emails

Below about 5,000 sends per version, the difference you are looking at is noise. Most people call a winner at 200 emails and then build a whole campaign on a coin flip.

Section 08

Handling replies

This is where most outbound quietly dies. The list and the email did their job, somebody answered, and then the answer sat for 3 days or got the wrong response back.

Reply in minutes, not days

The gap between someone replying and hearing back is the most visible quality signal you control, and it is the one thing a small operation can beat a big team on outright.

The 6 kinds of reply

ReplyWhat it looks likeWhat you send back
InterestedAny sign of interest, including a vague "sure" or "tell me more"A short explanation plus a booking link
A questionSomething specific they want to knowAnswer it, then the explanation, then the link
Wants to bookScheduling only, no questionJust the scheduling reply, nothing else
Out of officeAn automatic replyNothing. Wait for them to come back
A clear no"Not interested", "remove me"Nothing, and never email them again
HesitationA worry or a hurdle, not a refusalOne calm reply. No link, no ask

Judge the tone, not the words

Sarcasm, mockery, or a flat "stop" gets no reply at all, even when there is a question mark in it. Replying to sarcasm does lasting damage to how you are seen. Missing one borderline positive costs you a single reply. Those are not the same size of mistake.

Someone genuinely asking "is this real?" is the opposite, and deserves a straight answer. A cautious person is doing exactly what a careful buyer should do.

A question about money is a question, not a rejection

"How much is this?" arriving alongside interest is one of the best replies you will get. Treat it as an objection and you will strip out both the answer and the booking link, which is the worst possible response to the most important message in the thread.

Answer it plainly and keep the next step attached.

Write your standard answers once

You will answer the same 8 questions forever. Write each answer properly one time and reuse it word for word.

Rewriting from scratch every time feels more personal and is not. The wording drifts a little each round until you are saying something you never intended to promise.

The 4 hesitations, and how to answer them

Each answer is one or two sentences. Name the worry, take the air out of it, stop. No link, no ask attached. Adding a booking link to a reassurance turns it into a push and converts worse than saying nothing.

  • "What is the catch?" There is not one. Say what each side actually gets in one sentence and let how even it sounds answer the question.
  • "I am slammed." Give them the exact time cost. Vague reassurance loses badly to a precise number here.
  • "We are probably too small." Confirm they are exactly who you are looking for rather than arguing. Give them the real reason, not flattery.
  • "I am not sure about this." Name what stays under their control. Most hesitation is about losing control, not about the offer.
Part Three: After The Reply
Section 09

The call before the call

A yes from a cold email is fragile. Someone agreed to something in 4 seconds while clearing their inbox. They have not really decided anything yet, and a chunk of them will quietly not show up.

A short call in between fixes most of that. Book 15 minutes. It usually runs 8 to 10. It has exactly 2 jobs, and a call that does neither should not have happened.

Job one: agree on one topic

Come off the call with a single subject they will talk about with real confidence. One. Not a shortlist, not three angles to choose between on the day.

Here is how it goes wrong, and it goes wrong constantly. You offer 3 options to be accommodating. They say "honestly, all three". You say "perfect". Now nobody has chosen anything, and you both find that out live on the recording.

It feels like a win in the moment, which is exactly why it keeps happening. "All three" means they have not decided.

The fix is to choose for them, out loud, and check it back. "Let's build it around the second one then, and everything else can come in around it. Sound right?" They almost always agree, and now you have a topic.

Job two: set the date while you are still on the call

Agree the date together, before you hang up. Ending on "send me some times" or "reply with your availability" is a date that slips, and a date that slips is usually a date that never happens.

Two things you do the same day

  • Send the calendar invitation right after the call. Not that evening.
  • Send the topic outline by end of day. The thing you agreed, written down, so they can see it exists.

Both are small. Both prove you do what you say inside a few hours, which is most of what they are quietly assessing.

What to cover

  • That there is no cost to them.
  • Exactly what they walk away with.
  • Which platform you will be on.
  • That they should block 45 minutes, so nothing gets booked against it.
  • Who the audience is.
  • That notes will be taken.
Why this works

Someone who chose their own topic and said a date out loud has committed twice, in their own words. Someone who did neither is holding a slot in a calendar and nothing more.

We do not have a clean measured number for the difference, so we are not going to invent one. The reasoning is the argument.

Section 10

Running the recording

Book 45 minutes. It usually runs 35 to 40. Use whatever video software you already use, so nobody has to install anything. No slides, no script.

Talk less than 40 percent of the time

Go past that and you have quietly become the guest. Track it once and you will be surprised where you actually land.

Stay curious, not strategic

The moment you slip into sales voice, they feel it. The whole conversation changes register and they stop telling you things. Whatever you were about to gain by steering, you lose twice over.

Interrupt with something specific they said

When someone starts rambling, "interesting, tell me more" makes it worse. Cut in with a detail from 5 minutes ago instead. "Go back to the pricing change, you said it broke something."

Referring back proves you were actually listening, and that buys you the next 10 minutes of honesty.

Let silence sit

After a hard question, do not rescue them. The 3 seconds you desperately want to fill is where the real answer comes from. Almost everyone fills it and loses the best part of the conversation.

Spend the first stretch on easy ground

You need real rapport before a sharper question lands. Ask something pointed at minute 2 and you get a polished non answer, and they stay guarded for the rest of it.

Never pitch during the recording

Not softly, not at the end, not as a throwaway. This is the rule that matters most.

The recording is worth what it is worth precisely because it is not a sales call. One pitch turns it into one retrospectively, in their mind, including the 40 minutes before it. If there is a conversation to have about working together, it is a separate conversation on a different day.

What they get, and when

  • You own the recording.
  • It comes back properly edited.
  • They get a copy.

Promise a turnaround you can actually hold, and say it out loud on the prep call rather than leaving it vague. What limits you is editing capacity, not how fast one edit takes, so base the promise on how many recordings you book in a week.

Being a week late on something they were looking forward to undoes most of the goodwill the conversation just built.

Want this run for you instead?

Everything on this page is what we do for clients. If reading it makes you want to build it yourself, build it. If it makes you want it running by next month, book a call.

Book A Call

Common questions

How do I know whether my list or my writing is the problem?

Check two numbers once you have sent about 5,000 emails: how many people replied, and how many of those replies were positive. Under 1 in 100 replying, or under 3 in 10 of those being positive, means the list. Change who you are targeting before you touch the writing.

How many cold emails can one account send per day?

20. That is the ceiling after 14 days of warmup, built up gradually from 10 a day, then 15. If you want more volume, add more accounts rather than pushing one account harder.

How long should a cold email be?

65 words maximum, one sentence per paragraph. About 8 in 10 cold emails are read on a phone, and three sentences stacked together look like a wall of text the thumb scrolls straight past. Our best performing email was 38 words.

Why should I turn off open and link tracking?

Tracking rewrites your links to point at your email tool's shared tracking address first. If that address gets blacklisted, every email you send carries a blacklisted link. When we audited our own accounts, 123 of 258 were pointed at a blacklisted address and it affected nothing, because tracking was off and our emails contained no links.

Should I use a tool that guesses whether an unverified address is real?

It depends on company size. At companies of 1 to 20 people, guessed addresses bounced 2.5 percent of the time across 7,196 emails. At companies of 50 or more, the same guessing bounced 17.4 percent across 852 emails, against 3.0 percent for fully verified addresses. At larger companies, drop those addresses rather than guess.

What is the single biggest thing most people never test?

Who the email comes from. We ran identical emails under 2 senders and the founder version got 3.4 replies per 100 against 1.2 for a producer writing on his behalf. That is a bigger difference than any wording change we have ever tested.