---
title: "Cold Email Best Practices: 11 Rules From 1.2M Emails We Sent"
url: https://sales.co/research/cold-email-best-practices
source: https://sales.co/research/cold-email-best-practices
description: "11 rules from 1,177,370 cold emails Sales.co sent. Offering to send something got 46% more replies than asking for a call; a first name in the subject, 49% more."
---

# Cold email best practices: 11 rules from 1,177,370 emails we sent

These cold email best practices held in 1,177,370 first emails Sales.co sent from Oct 6, 2025 to Nov 1, 2026. Offer to send something instead of asking for a call (0.77% vs 0.53% human replies per email, across 565,286 and 158,310 emails). Put their first name in the subject and keep it to 5 to 7 words. Greet them by name, and write one or two paragraphs with one question. Each of these 11 rules got at least 10% more human replies than breaking it, and kept that lead with our largest sender removed. The 3 subject line rules come from 730,888 first emails in the same window.

By [Ryan Doyle](https://www.linkedin.com/in/ryan-doyle/), co-founder of Sales.co · Updated November 16, 2026 · Data through November 1, 2026

1,177,370

first emails, Oct 6, 2025 to Nov 1, 2026

8,336

human replies, 0.71% of emails

58

sending workspaces; the largest sent 17% of emails

## The 11 rules

The number is how many more human replies per email the rule got, against breaking it.

1. 01Offer to send something. Don't ask for a call.+46%
2. 02Put their first name in the subject.+49%
3. 03Write a 5 to 7 word subject.+88%
4. 04Capitalize the first letter of the subject.+44%
5. 05Greet them by first name.+47%
6. 06Say who you are and what you do.+16%
7. 07Keep it to one or two paragraphs.+46%
8. 08Ask exactly one question.+27%
9. 09Leave out the stats.+10%
10. 10Cut the exclamation marks.+32%
11. 11Add a P.S.+29%

## Each rule, with the numbers

Black bar: emails that followed the rule. Grey bar: emails that broke it. Every rule was checked again with our largest sender removed.

01

### Offer to send something. Don't ask for a call.

End on an offer: a short video, an example, a list. Of the 4 endings we measured, asking for a call got the fewest human replies and the fewest interested replies.

DoWant me to send over the 2-minute walkthrough?

Don't

Do you have time for a quick call this week?

+46%

human replies per email

Offers to send something0.77%

Asks for a call0.53%

Largest sender removed

+36%

Human replies counted

4,374 vs 840

Interested replies

0.25% vs 0.10% (1,424 vs 161)

Emails compared

565,286 vs 158,310

02

### Put their first name in the subject.

A subject with their name in it reads like it was written to one person, because it was.

DoSarah, idea for Acme's evening calls

Don't

Idea for Acme's evening calls

+49%

human replies per email

First name in subject0.83%

No first name0.56%

Largest sender removed

+29%

Human replies counted

1,731 vs 2,696

Interested replies

0.30% vs 0.19% (622 vs 917)

Emails compared

208,847 vs 485,776

03

### Write a 5 to 7 word subject.

One and two word subjects were the worst length we sent. A subject long enough to say what the email is about did best.

DoSarah, idea for Acme's evening calls

Don't

Quick question

+88%

human replies per email

5-7 words0.79%

1-2 words0.42%

Largest sender removed

+99%

Human replies counted

316 vs 331

Interested replies

0.35% vs 0.14% (138 vs 113)

Emails compared

39,841 vs 78,463

04

### Capitalize the first letter of the subject.

The all-lowercase subject is a known cold email trick, and it got fewer replies than a normal sentence.

DoQuick idea for Acme

Don't

quick idea for acme

+44%

human replies per email

First letter capitalized0.67%

All lowercase0.47%

Largest sender removed

+25%

Human replies counted

2,199 vs 262

Interested replies

0.22% vs 0.14% (714 vs 76)

Emails compared

326,372 vs 56,132

05

### Greet them by first name.

"Hi there," reads like a list. "Hi Sarah," reads like one person wrote to another.

DoHi Sarah,

Don't

Hi there,

+47%

human replies per email

"Hi Sarah,"0.81%

Greeting without a name0.55%

Largest sender removed

+38%

Human replies counted

5,315 vs 2,428

Interested replies

0.26% vs 0.13% (1,681 vs 571)

Emails compared

652,574 vs 438,691

06

### Say who you are and what you do.

We counted pronouns. Emails that used "I" and "we" more than "you" got more replies than emails built around "you". In practice that means saying who is writing and what you do.

DoI'm Ryan from Fieldline. We schedule service techs for HVAC companies.

Don't

You're probably losing jobs after 5pm, and your team...

+16%

human replies per email

More "I/we" than "you"0.75%

More "you" than "I/we"0.65%

Largest sender removed

+22%

Human replies counted

4,058 vs 2,786

Interested replies

0.23% vs 0.17% (1,238 vs 737)

Emails compared

538,231 vs 427,834

07

### Keep it to one or two paragraphs.

Word count made no clear difference once the largest sender was removed. The number of paragraphs the reader has to get through did.

DoTwo short paragraphs and a sign-off.

Don't

Seven one-line paragraphs stacked down the screen.

+46%

human replies per email

1-2 paragraphs0.84%

7 or more0.58%

Largest sender removed

+18%

Human replies counted

495 vs 137

Interested replies

0.27% vs 0.13% (156 vs 31)

Emails compared

58,650 vs 23,648

08

### Ask exactly one question.

An email with nothing to answer got the fewest replies. Two or more questions made no clear difference against one.

DoWant me to send it?

Don't

Let me know if this is of interest.

+27%

human replies per email

One question0.71%

No question0.56%

Largest sender removed

+19%

Human replies counted

6,070 vs 452

Interested replies

0.21% vs 0.13% (1,805 vs 107)

Emails compared

853,201 vs 80,498

09

### Leave out the stats.

Emails with a "40% more pipeline" or "3x faster" claim got fewer replies, and the gap in interested replies was wider.

DoTwo HVAC shops your size use it for evening calls.

Don't

Our clients book 3x more jobs.

+10%

human replies per email

No % or Nx figure0.72%

A % or Nx figure0.66%

Largest sender removed

+17%

Human replies counted

6,653 vs 1,683

Interested replies

0.22% vs 0.16% (2,000 vs 418)

Emails compared

920,253 vs 257,117

10

### Cut the exclamation marks.

Excitement from a stranger reads as a sales pitch.

DoCongrats on the new Tampa location.

Don't

Congrats on the new Tampa location!

+32%

human replies per email

No exclamation mark0.71%

One or more0.54%

Largest sender removed

+18%

Human replies counted

8,145 vs 191

Interested replies

0.21% vs 0.10% (2,384 vs 34)

Emails compared

1,142,162 vs 35,208

11

### Add a P.S.

The P.S. is the line people read after skimming the rest. Use it for something true and personal that asks for nothing.

DoP.S. Saw the new Tampa location. Congrats.

Don't

(nothing after the sign-off)

+29%

human replies per email

Has a P.S.0.89%

No P.S.0.69%

Largest sender removed

+24%

Human replies counted

728 vs 7,608

Interested replies

0.28% vs 0.20% (231 vs 2,187)

Emails compared

81,469 vs 1,095,901

## An email that follows all 11

Written for this page, for a made-up HVAC scheduling company.

Subject: Sarah, idea for Acme's evening calls

Hi Sarah,

I'm Ryan from Fieldline. We schedule service techs for HVAC companies, and I noticed Acme's booking line closes at 5pm.

I put together a one-page breakdown of how two shops your size book evening calls without hiring. Want me to send it?

Ryan

P.S. Saw the new Tampa location. Congrats.

## Advice that made no clear difference

Common cold email rules that didn't move replies by 10% both overall and with the largest sender removed. Shown as overall / largest sender removed.

- Keep it to 25-50 words+19% / -1%
- Write at a 4th-5th grade reading level+3% / -36%
- One sentence per paragraph-6% / +4%
- Tell a customer story+2% / +3%
- Mention their company+9% / +8%
- Cut "hope you're well"-4% / +4%
- Avoid hedge words (might, maybe)-3% / -12%
- Write in an informal tone+24% / +8%
- Ask a question in the subject+9% / -5%

## Limits and the strongest objection

This is observational data, so it shows which emails got more replies and does not prove the trait caused them. Different senders write different emails to different lists, so a trait can look good because the senders who use it have better lists. Two checks limit that: every rule is checked again with the largest sender removed (it sent 17% of the emails), and no group is shown when one sender made more than 45% of it.

The rules also overlap. An email that follows one often follows others, so the gaps don't add up, and part of each gap may belong to the rest. These are Sales.co's own campaigns, with our clients' lists and offers; a different mix of industries and offers may show different gaps. Treat each rule as a strong bet for your own split test.

For reply rates by word count, see [how long a cold email should be](https://sales.co/research/how-long-should-a-cold-email-be); for reply rates by what the last line asks for, the [cold email call to action](https://sales.co/research/cold-email-call-to-action) study; for the greeting and first line, [how to start a cold email](https://sales.co/research/how-to-start-a-cold-email). For reply rates across all our sends, see the [cold email statistics](https://sales.co/research/cold-email-statistics), and for when to send, the [best time to send cold email](https://sales.co/research/best-time-to-send-cold-email). To check a subject line against these rules, use the [subject line grader](https://sales.co/cold-email-subject-line-grader); to check a draft for spam trigger words, the [email spam checker](https://sales.co/email-spam-checker).

## Method and FAQ

Cite as: Sales.co, "Cold email best practices: 11 rules from 1,177,370 emails we sent", https://sales.co/research/cold-email-best-practices, data through November 1, 2026.

How the data was collected Source: the text of every cold email Sales.co sent from October 6, 2025 to November 1, 2026 (1,325,479 first emails scanned), and the replies the sending tool labeled. Unit: recipient thread: one recipient address in one sending workspace, described by the first email that address got. 1,177,370 recipient threads from 58 sending workspaces. Subject line rules use 730,888 threads (4,653 human replies, 52 sending workspaces). A human reply is the recipient replied at least once with a reply that is not an auto-reply, out-of-office, bounce, warm-up, test or spam (sending tool's labels). An interested reply means at least one of those replies was labeled 'possibly interested'. A script turned each email body into traits (paragraphs, questions, the ask in the last paragraph, stats, greeting, P.S.) and dropped the text. No email, subject line, address or client is published. A rule is listed when following it got at least 10% more human replies per email than breaking it, overall and with the largest sender removed, with 10,000+ emails on each side. Left out: any group with fewer than 2,000 emails, or where one sending workspace made more than 45% of them. The newest 14 days are left out so replies can arrive. The page is regenerated from the data each time it is recomputed, so a rule that stops holding drops off.

What gets cold emails the most replies? In 1,177,370 first emails Sales.co sent, the widest gap in interested replies among the 11 rules came from the ask. The 565,286 emails that offered to send something (a video, a list, an example) got 0.77% human replies per email, against 0.53% for the 158,310 that asked for a call or meeting, and 2.5x the interested replies.

Should you ask for a meeting in a cold email? Not in the first email. Asking for a call got 0.53% human replies and 0.10% interested replies per email. Offering to send something got 0.77% and 0.25%. The gap held with the largest sender removed (0.69% vs 0.51%).

How long should a cold email subject line be? 5 to 7 words did best: 0.79% human replies per email, against 0.42% for 1 to 2 word subjects, out of 730,888 first emails. Subjects with the recipient's first name got 0.83% against 0.56% without it.

How long should a cold email be? Word count made no clear difference once the largest sender was removed. Paragraph count did: emails with 1 to 2 paragraphs got 0.84% human replies per email, against 0.58% for 7 or more.

Do statistics help a cold email? No. Emails with a percentage or "Nx" figure in the body got 0.66% human replies and 0.16% interested replies per email, against 0.72% and 0.22% without one.

Should a cold email have a P.S.? Yes. Emails with a P.S. got 0.89% human replies per email against 0.69% without one, and 0.77% vs 0.63% with the largest sender removed.

## Sources, as of November 16, 2026

- Sales.co sending database, email text: 1,177,370 first cold emails and 8,336 human replies from 58 sending workspaces, October 6, 2025 to November 1, 2026, as of November 16, 2026.
- Sales.co sending database, subject lines: 730,888 first cold emails and 4,653 human replies from 52 sending workspaces, May 1, 2026 to November 2, 2026, as of November 17, 2026.
- Sales.co pricing and plans: [sales.co/pricing](https://sales.co/pricing), read October 25, 2026.

Sales.co writes and sends cold email like this for B2B teams, from $2,000 a month for 10,000 emails.

[See pricing →](https://sales.co/pricing)
