Skip to content
From the blog

Cold email opening lines that earn the read

A cold email opening line has one job: to make the second sentence worth reading. Almost every opener in circulation fails it the same three ways, and nobody has measured any of them, because no dataset on email first lines exists. Here is the catalog, the test, and the evidence that does exist.

By Kshitij Maheshwari, co-founder · Updated August 2026 · 11 min read

The job

The first line has one job

Our cold email copywriting guide settles what an opener is for. What it does not say is where the first line is read, which is not inside the email.

Definition

An opening line is the first sentence of the body, the one line every reader sees. In a plain text email with no preheader, it is also your preview text.

Litmus documents that clients typically build that preview from the first lines of copy, and recommends staying under 90 characters because you cannot tell how much will show (updated November 2024).

Your opener is read in the inbox list first. By the time it is sentence one, it has already worked or it has not.


The catalog

Every opener people send, and whether it earns line two

Thirteen patterns, judged on one question rather than on whether they sound good. Shapes, not lines: a line that ranks on page one is a line your reader already knows.

Pattern What it sounds like Earns line two? Why
The compliment Really enjoyed your talk last month No Asks a stranger for warmth before they know what you want.
The trivia line Saw you also went to Michigan No About the person, not their week. Nothing for them to tell you.
The mutual connection We are both connected to Priya Sometimes Only if you would be happy for Priya to read it forwarded.
The referral Priya suggested I write to you Yes, when true A borrowed warrant, checkable in one message. Never stretch it.
The question Who ended up owning renewals after the reorg? Yes, if only they can answer it The strongest family and the most misused.
The provocation Most teams buy the data tool before they agree on the account list Yes, carefully Provoke the practice, never the person.
The statistic 87% of B2B teams waste their first quarter No Homework in sentence one, and it says nothing about them.
Quick question Quick question for you No Announces a question instead of asking one.
The permission ask Mind if I ask you something? No Buys a turn the reader granted by opening the message.
The introduction I am Dana, I run partnerships at Corvid No Repeats the From line they have already read.
The null opener We help companies like yours... No Describes a category, which is not a fact about them.
The flat observation You closed the Series A in March and the first ops hire started last week Yes True of them and few others, and the implication is left open.
The direct opener I am writing because you own onboarding at a company that just doubled its sales team Yes, with a ceiling Costs nothing and buys honesty. Sentence two carries everything.

Sources: no vendor or study has published a measured dataset on cold email opening lines, so this table is a taxonomy rather than a ranking. The permission ask is the one pattern with data against it, from Gong Labs' 90,380 cold calls (2018), a different channel. Preview behavior: Litmus, November 2024.


The test

The line-two test

Three checks, about ten seconds, run on a sentence you have already written. Any one failure is fatal on its own.

Check one
Predict it

Cover everything but the first sentence and say what sentence two will be. Get it right and so will they, and a correct guess is as good as reading.

Check two
Own it

Find the grammatical subject. If it is I or we, you spent the one guaranteed line narrating your afternoon instead of their week.

Check three
Answer it

After sentence one, is there anything the reader knows that you do not? A line that closes its own loop has left them nothing to do.

The test judges the line, not the belief behind it. Which reasons a stranger has to reply is a separate question.

The first line does not have to be interesting. It has to be unpredictable.


The receipts

Stop trying to look important

Nobody has measured email first lines, so the two findings worth borrowing come from next door. Both point the same way: the reader answers a question, not an announcement.

Two findings, borrowed
Looking important

Ask something only they can answer

Signaling that your email is important barely moves replies; asking something only they can answer does. That is a 2005 Carnegie Mellon study of real inboxes, and it measured colleagues rather than strangers.

Asking permission

Never ask permission to ask

Gong Labs' 2018 analysis of 90,380 cold calls: asking whether you caught them at a bad time booked meetings at 0.9% against a 1.5% baseline. Vendor data, telephones not inboxes, and Gong's own page says openers have a shelf life.

Gong's top call opener is asking how someone has been, at 10.01%. That is not a case for rapport in email: a spoken opener compels a turn, an inbox compels nothing, and a claim of prior contact is one an email reader disproves by scrolling.

Operator note
What I check first

I stopped asking whether a first line was good and started asking whether it was answerable. Good is a taste argument you lose with yourself at midnight. Answerable is a property of the sentence, and you can check it in one pass.

KM
Kshitij Maheshwari
Co-founder, Real Good GTM

The vacuum

Nobody has measured this, including us

Every confident number about email first lines is either unmeasured or borrowed from an aggregator. That absence is worth more to you than another list of lines.

What the category claims

"Hope all is well lifts meetings 24%, personalized openers reply at 17%, and you have six seconds to hook them."

  • The 24% sits on a vendor page with no sample and no success definition
  • The 17% traces to an aggregator that states no collection period
  • The six seconds names no study anywhere in its chain
What actually exists

"One opener dataset in outbound discloses its sample, and it is about telephones."

  • Gong has published on email length, ROI language and asks, never openers
  • lemlist, Instantly and Smartlead have published nothing on first lines
  • The one disclosed set, 90,380 calls, is a different channel
Where that leaves us

We have no number for you, and neither does anybody else. What we have is a shape that survives the reader guessing what comes next, and a check you can run on your own sentence in ten seconds.


The construction

How to write a flat observation

The answer everyone prescribes and nobody builds. Four rules, in the order you apply them to a draft.

  1. 1

    Make the reader the subject

    You are the first ops hire, rather than I noticed you are the first ops hire. The noticing is not the news, and deleting I saw that is usually the whole edit.

  2. 2

    Use a fact they can confirm from memory

    If they would have to look it up, it is research you did rather than a situation they are in. Finding that fact is its own discipline, and it is not the same job as writing the line.

  3. 3

    Delete every adjective and see what is left

    Impressive growth, exciting round, big move. If the sentence stops saying anything, there was never an observation, only a compliment with a fact attached.

  4. 4

    Leave the implication open

    A fact with nothing after it is surveillance with a full stop. It has to lead somewhere they can agree with or correct, which is what turns a fact into a question.

Want the two of us writing your first lines, on real signals?

Book a Fit Check

The rewrites

Three openers, rewritten

An illustrative walkthrough of the method, not a specific client result. We report real numbers only when they are real. The companies and people are invented, and no reply rate is attached to any of these.

Don't

Open with your admiration

Really enjoyed your talk on RevOps hiring, it resonated.

  • Sentence two is a pitch and they know it
  • The subject of the sentence is you
  • Nothing to confirm or correct
Do

Open with what they said

You are hiring a RevOps lead before there is a RevOps system, which is the order most teams end up in.

  • Their words, not your reaction to them
  • A read they can agree with or correct
  • Sentence two is not guessable
Don't

Open with a number

87% of Series A companies waste their first quarter on the wrong ICP.

  • Homework before they know why you wrote
  • True of a category, not of them
  • A stranger's figure is a claim to evaluate
Do

Open with two facts of theirs

Your two newest AE hires both came from enterprise, and pricing still starts at 25 seats.

  • Both checkable from memory in a second
  • One tension, left implied
  • Nothing in it to disbelieve
Don't

Ask before you ask

Quick question, mind if I ask what you use for enrichment?

  • Announces the question instead of asking
  • Buys permission the open already granted
  • Spends the one guaranteed line
Do

Ask the question

You are running Clay alongside two enrichment vendors, which usually means paying twice for the same records.

  • The question is implicit and specific
  • Only they can say whether it is true
  • No line spent on preamble

The ops

How to tell whether an opener worked, with no data

You cannot measure it. Nothing in a sending stack records whether anyone read your second sentence, so here are four instruments that work at founder volume.

Four instruments
The replies

Read the reply, not the rate

At twenty replies you have no statistics and a great deal of information. A reply that quotes your first line back means it landed. Who is this means it never established why you wrote.

The mix

Sort replies by kind, not by count

On topic against confused against annoyed. That mix moves visibly when the opener changes, long before any rate does.

The preview

Cut it at 90 characters and read it cold

Paste the first line into a document, cut it where the inbox will, and read what survives. That is what the reader sees before deciding anything.

The portfolio

Sort your last thirty first lines by pattern

If twenty-five are the same shape, you do not have thirty data points. You have one, sent thirty times.

Two things cannot referee a first line: open rate, which stopped being a measurement after Apple's Mail Privacy Protection, and an A/B test at founder volume, where the sample arithmetic does not work.

Operator note
At founder volume

Twenty replies will not give you a rate, and they will tell you exactly what your first line is doing. We read them before we read any dashboard, because a dashboard cannot tell confused apart from uninterested.

RB
Rahul Bageria
Co-founder, Real Good GTM

Failure modes

Five ways a good opener still dies

A line can pass all three checks and still fail, because most of these are about everything around the sentence.

The clock ran out

The event that made the observation true stops being news, and the same sentence six weeks late reads as scraping. Which windows close when is the trigger event play.

A portfolio of one shape

Thirty first lines with the same structure are one experiment run thirty times. Whatever the reply rate is, it is a fact about that shape, not about your market.

The observation with no implication

A true fact, stated flatly, and then nothing. The reader has been noticed at and handed nothing to answer.

A good line on the wrong offer

The opener earned sentence two, and sentence four had nothing they wanted. No first line rescues a message written to a list nobody had a reason to write to.

!
Caution

The referral you stretched

Every other bad opener costs you one email. A name you did not quite earn costs two relationships, because the reader can check it in one message and often does.

Do this instead
Never write a name you would not want forwarded, and assume it will be.

The short version

Key takeaways

Four things to keep after you close this page.

Key takeaways
4 points
  • 1 Your first line is read in the inbox list before it is read in the email.
  • 2 If you can predict sentence two, so can they, and a guess is as good as reading.
  • 3 Openers are written to prove importance. Answerability is what moves replies.
  • 4 Nobody has measured email first lines, so judge the shape, not the rate.

FAQ

Questions people ask

What is a good opening line for a cold email?
One the reader could not have received from anybody else that week. Their situation, stated flatly, with them as the subject of the sentence and the implication left open so they can agree with it or correct it. Both of those are replies.
What should you never say in the first line of a cold email?
Anything that announces the second sentence. Quick question, really enjoyed your talk, and I am Dana from Corvid all tell the reader what is coming next. A reader who can predict the next line has no reason to read it.
Should you open with hope you are doing well?
There is no evidence either way for email. The 24% figure people quote sits on a vendor page with no sample size and no definition of success for that finding, unlike other studies on the same site that disclose both. Treat it as unmeasured, and note that it fails the prediction test anyway.
Should you compliment someone in a cold email?
No. A compliment announces sentence two, and whether to read sentence two is the only thing your first line decides. A 2010 marketing experiment did find that flattery buys warmth even when the reader sees the motive coming, but warmth toward you is not a reason to keep reading.
Should you open with a question?
Yes, if the reader is the only person who could answer it. A 2005 Carnegie Mellon study of real inboxes found information requests were 22% more likely to get a reply. Are you looking to improve pipeline is answerable by anyone, which is why nobody answers it.
Is it better to be direct or to build rapport first?
Direct, in email. Gong's analysis of 90,380 cold calls put rapport openers on top, but a spoken opener compels a response and an inbox compels nothing. A line claiming prior contact is also one an email reader disproves by scrolling.
Kshitij Maheshwari, co-founder of Real Good GTM
About the author
Kshitij Maheshwari

Co-founder of Real Good GTM. He has been the first business hire and Chief of Staff at seed-stage B2B startups, building outbound pipeline before any playbook existed. This post comes from writing first lines every week, and from tracing each opener statistic back to whoever published it.

Connect on LinkedIn

Keep going

The rest of the message

The first line is one of five parts. Here are the other two that decide whether it gets read.

Want first lines nobody else could have written?

Book a fit check. We'll look at what your market actually emits, which observations are worth writing about, and tell you straight if outbound is not the right motion for you yet.

Book a Fit Check

No hard sell. No fake numbers. Real good work speaks for itself.