Do cold emails still work in 2026?

Yes, at lower volume and higher effort than in 2018. Inbox filtering got stricter, Google and Microsoft tightened bulk-sender rules, and buyers have seen every AI-written opener. What still works is a small list you chose deliberately, one researched line per email, and thirty to forty sends a day from a warmed domain. What stopped working is five thousand contacts and a spun first paragraph.

How many follow-ups should I send?

Two, then a break-up. That is four emails across twelve days, all in the same thread, each one adding something the previous email did not say. Sequences of seven and nine still exist and they still technically produce meetings, but the damage to your domain reputation and to the way people talk about you is real, and a founder cannot afford either.

What's a good cold email reply rate?

For a founder sending a hand-picked list in B2B SaaS, five to ten percent total replies is a working campaign and one to three percent positive replies is normal. Under two percent total, stop tuning the copy and change the list or the offer. Above fifteen percent, your list is probably warm rather than cold, which is good news you should learn from.

How long should a cold email be?

Between fifty and ninety words for a first touch, and under forty for a follow-up. The real test is whether the whole thing fits on a phone screen without scrolling, because that is where it will be read and where the decision to reply or archive is made.

Should I use AI to write cold emails?

Use it for the structure and the boring middle, never for the researched line. The opening sentence is the only part carrying proof that a person looked at their company, and a model that has not seen their job ads or their changelog cannot write it. Drafting with AI and researching by hand is the combination that holds up.

In the US, CAN-SPAM permits B2B cold email if you identify yourself, give a real postal address, and honour opt-outs. In the UK and EU, business-to-business email to a corporate address is generally permitted under a legitimate-interest basis, with a clear identity and an easy opt-out, though rules differ by country and are stricter for sole traders and partnerships. Check your own jurisdiction before you scale, and always honour the first unsubscribe.

← All posts

Cold

Cold email templates for SaaS founders: 13 that still get replies

8 August 2026 · 12 min read

Thirteen cold email templates, each one short enough to read on a lock screen and plain enough to send from your own Gmail. Every one has the situation it is for, the subject line, the body you can copy, and the single line that people ruin it with.

Most swipe files fail founders for the same reason. They were written by an agency running four thousand sends a month for a client with a sales team, so the templates quietly assume a sequencer and somebody whose entire job is that inbox. You have ninety minutes before the support queue opens. A cold email template a startup can actually use has to survive that, so the ones below are sized for it.

One bit of context first, in case you have never seen it laid out. There are only four things you can do to get a customer. Reach out to people who already know you. Post something in public where strangers can find it. Reach out to strangers. Pay to get in front of people. Alex Hormozi calls that set the Core 4 in his book $100M Leads, and this page is about the third one, cold outreach.

A warning before the swipe file. Templates are the fifth most important thing in cold outreach, behind the list, the offer, the timing, and the volume. A brilliant email to the wrong two hundred people loses to an average email to the right forty. If you have not settled on cold outreach yet, read how the Core 4 actually works first, or take the two-minute channel quiz. Cold is the most expensive of the four to run badly.

Five rules that make any template work

Skip these and the templates below will do worse than a blank page. A great template sent to a bad list loses to a clumsy email sent to the right people. None of this takes long to set up. All of it decides whether the next two hundred sends teach you anything at all.

01

One researched line, ninety seconds to find it

One sentence that could not have been sent to anyone else. Ninety seconds is the cap. A job ad, a changelog entry, a number on their pricing page, something they said on a podcast. If it takes you fifteen minutes to find, you have built a channel you can run forty times and then never again.
02

Cap the volume before you touch the copy

Thirty to forty sends per mailbox per day, from a separate sending domain, warmed for two weeks first. Under about fifty sends of one version, a reply rate is noise, and you will read the wrong thing into it. Work out how many emails you need before you write any, because the number changes what you write.
03

One ask, and it appears once

A reply, a call, a video, a forward. Pick one. Two asks in one email reliably get you zero, because choosing is work and the reader is on a phone in a queue. Put the ask on its own line at the bottom. Never repeat it in a postscript.
04

Subject lines: two to five words, under forty characters

Write the subject a colleague would write. Lowercase, no claim, no emoji, no stacked question marks. "quick question, {{company}}" beats anything with a benefit in it, because a benefit in the subject line is the tell that this is a marketing email. Over about forty characters it gets cut off on a phone anyway.
05

Know the two moments to stop

Stop on a person after four emails across twelve days, the last of them a real break-up, then leave them alone for ninety days. Stop on cold outreach altogether if two hundred sends across two different offers have not produced a two percent reply rate. At that point the list or the offer is wrong, and better copy fixes neither.

Rule two is the one founders skip, and it is the one that decides the outcome. Forty a day for six weeks is roughly seventeen hundred emails, which at a three percent reply rate is fifty conversations. Twelve a day gives you fifteen conversations over the same six weeks, and fifteen conversations cannot tell you whether cold outreach works for you. The volume estimator does that sum in about twenty seconds. It is free.

The 13 templates

Swap anything inside double braces. Almost every merge field here comes straight from a spreadsheet column. The one that does not is {{observation}}, the researched line, and that is the whole reason these B2B cold email examples get answered while the same structure sent blind does not.

First touch to a named prospect

Use it when: You found one company on purpose, you know what they do, and nobody has introduced you.

Subject quick question, {{company}}

Hi {{first_name}},

{{observation}}. That usually means {{consequence}} somewhere behind it.

We make {{my_product}} for exactly that. {{peer_company}} used it and {{outcome}}.

Worth a look, or is this the wrong month?

{{my_name}}

Why it works: It proves you looked, offers one comparable customer, and hands them an easy no that costs them four words.
The line most people get wrong: The "we make" line. Most people turn it into three sentences of positioning. One sentence, one outcome, or the whole email gets skimmed and binned.

The referral ask

Use it when: Someone you already know is one step from the person you want. This is the highest-converting email on the page and the one founders send least.

Subject intro to {{contact_name}}?

Hi {{first_name}},

You mentioned you know {{contact_name}} at {{company}}. Would you be up for an intro?

Here is something forwardable: we make {{my_product}}, and {{peer_company}} used it to {{outcome}}. I want fifteen minutes to find out whether it is relevant to them. That is the whole ask.

Completely fine to say no, it will not be weird.

{{my_name}}

Why it works: The forwardable paragraph does the referrer's work for them, and the explicit permission to decline drops the social cost of asking to near zero.
The line most people get wrong: Ending with "let me know if you think it makes sense". That leaves your friend to compose the intro from scratch, which is why the intro never arrives. Write the forwardable text yourself.

The "I saw you are hiring for X" trigger

Use it when: They have a live job ad that implies the exact problem you solve. Dated, public, budget-backed evidence.

Subject hiring a {{job_title}}

Hi {{first_name}},

You have a {{job_title}} role open. In my experience that means {{observation}} is already painful, and the hire is three months from helping.

{{my_product}} covers part of that from week one, so whoever you hire starts on the interesting half. {{peer_company}} did it in that order and {{outcome}}.

Want the two-minute version?

{{my_name}}

Why it works: Timing beats copy. A company that is spending sixty thousand on a problem has already admitted the problem exists, so you skip the entire convincing stage.
The line most people get wrong: Opening with "congrats on the growth!". Nobody feels congratulated for posting a job ad. Say what the role implies instead, and be specific enough to be wrong.

The competitor switcher

Use it when: You can see they use a competitor, from a badge, a review, a job ad, or their integrations page.

Subject {{competitor}} question

Hi {{first_name}},

You are on {{competitor}}, so I will skip the comparison. One question instead: {{observation}}?

That is the thing people move to us for. If it is not a problem for you, ignore this and I will not chase it.

If it is, I will send a ninety-second video of how we handle it. No call needed.

{{my_name}}

Why it works: A diagnostic question respects that they chose the competitor for a reason, and the video is a smaller ask than a meeting, so more people take it.
The line most people get wrong: The feature comparison. Listing where you win assumes you know which trade-off they made deliberately, and when you guess wrong you have told a happy customer you did not understand their choice.

After a webinar, podcast or conference talk

Use it when: They appeared somewhere and you actually consumed it. This one does not work if you skimmed the show notes.

Subject your {{topic}} answer on {{show_name}}

Hi {{first_name}},

Listened to your {{show_name}} episode. The bit about {{topic}}, roughly thirty minutes in, was the first honest answer I have heard on it.

We are working the same problem from the other side with {{my_product}}. Not pitching you. I would like your read on one thing I think we have got wrong.

Fifteen minutes in the next two weeks?

{{my_name}}

Why it works: It asks for an opinion rather than attention, and people who agree to go on podcasts are people who like being asked what they think.
The line most people get wrong: "Loved the episode!" with nothing after it. If you cannot name the moment and roughly where it sat in the recording, they know you did not listen, and now the rest of the email is suspect too.

Reopening a stalled thread

Use it when: You had a real conversation, it went quiet somewhere between two months and a year ago, and nothing was ever decided.

Subject picking this back up

Hi {{first_name}},

We talked in {{month}} about {{topic}} and it went quiet. Usually that is me catching someone in a bad week.

Two things have changed since: {{change_one}}, and {{change_two}}.

Is this live again for {{quarter}}, or properly dead? Either answer is useful to me.

{{my_name}}

Why it works: Taking the blame for the silence removes the small embarrassment that stops people replying to an old thread, and naming two changes gives an honest reason to reopen it.
The line most people get wrong: "Just checking in." It carries no new information, so the reader learns that replying to you produces nothing. If genuinely nothing has changed, do not send this yet.

Follow-up one, three days later

Use it when: No reply to the first email. Send it as a reply in the same thread so the original is right underneath.

Subject (reply in the same thread, no new subject)

Hi {{first_name}},

One thing I should have led with: {{peer_company}} were {{observation}} before they started, and {{outcome}} inside six weeks.

Still very happy to hear a no.

{{my_name}}

Why it works: It adds a fact instead of repeating a request, which means it would still make sense as a standalone email. That is the test every follow-up should pass.
The line most people get wrong: "Bumping this to the top of your inbox." It announces that you have nothing new to say and that you are willing to send emails anyway.

Follow-up two, six days after that one

Use it when: Still nothing. This is the last one before the break-up, and it changes the question completely.

Subject (same thread)

{{first_name}}, is this you, or does someone else look after {{topic}} at {{company}}?

Happy to go away, equally happy to talk to them instead.

{{my_name}}

Why it works: A routing question can be answered in four words, and it gets forwarded internally far more often than a pitch does, which sometimes lands you the right person for free.
The line most people get wrong: Apologising for emailing again. An apology invites the reader to agree that you have been a nuisance, and you have handed them the frame.

The break-up email

Use it when: Three days after follow-up two. Twelve days and three emails in, still nothing. Send this and mean it.

Subject closing the loop

Hi {{first_name}},

I will stop here. No hard feelings and no more emails from me on this.

If {{observation}} becomes a priority later, reply to this thread and I will pick it straight back up.

Leaving the one-pager in case it is useful to whoever ends up owning it: {{link}}

{{my_name}}

Why it works: A real ending with a re-entry point gets a surprising share of the replies in any sequence, usually some version of "bad timing, ask me in the autumn", which is a genuine result.
The line most people get wrong: Manufactured finality, the "I will assume you are not interested and close your file" line. Readers have seen it four hundred times, they know your file does not exist, and it burns a contact you could have kept.

Asking for a fifteen-minute call

Use it when: Your product needs to be seen to be understood, or the deal is big enough that a call is the only honest next step.

Subject 15 minutes, {{topic}}

Hi {{first_name}},

{{observation}}. We built {{my_product}} for that, and {{peer_company}} got {{outcome}}.

Fifteen minutes Tuesday or Thursday to work out whether it applies to {{company}}? If it does not, I will tell you on the call and give you the time back.

{{my_name}}

Why it works: Naming the length, offering two days rather than an open calendar, and pre-committing to disqualifying yourself all reduce the perceived risk of saying yes.
The line most people get wrong: Dropping a booking link into a first email. The link is a second ask on top of the first, and asking a stranger to open your calendar before they have replied to anything is a bigger jump than it looks from your side.

Asking for a reply instead of a call

Use it when: Early in a campaign, or any time you care more about learning who has the problem than about booking this week.

Subject one question

Hi {{first_name}},

Quick one, no meeting: how does {{company}} handle {{topic}} today?

Asking because we built {{my_product}} for it, and I want to know whether our version is worth showing you or not.

One line back is plenty.

{{my_name}}

Why it works: The ask costs about eight seconds, so it clears the bar that a meeting request does not, and any reply gives you a live thread to work in.
The line most people get wrong: Burying the question under three lines of context. The question goes in the first or second line. Context earns its place only after somebody has agreed to keep reading.

The founder-to-founder note

Use it when: You are genuinely small and you are writing to somebody who is also genuinely small. Only send this if both halves are true.

Subject {{company}}, from a two-person team

Hi {{first_name}},

I am one of two people building {{my_product}}. There is no sales team here. I picked {{company}} off a list of {{observation}} and wrote this myself.

What it does, in one line: {{outcome}}.

If that is nothing to you, say so and I will take you off the list personally, today.

If it is something, I can show you it this week.

{{my_name}}

Why it works: Being small is an advantage in this one format, and the promise to remove them personally is a promise only a founder can credibly make.
The line most people get wrong: Wearing it as a costume. A team of twelve sending "I am one of two people" gets caught, because the reader checks the site, and the credibility you lose is worth more than the reply you were chasing.

To someone who has ignored you twice

Use it when: Two emails, no reply, and you are about to send a third. Send this one instead.

Subject last one, different angle

{{first_name}}, two emails from me and no reply. That is a clear enough answer on the pitch.

Different question, and it is genuinely useful to me: is {{topic}} not a problem at {{company}}, or is it a problem you solved a while ago?

Whichever it is, this is the last email either way.

{{my_name}}

Why it works: Dropping the sale and asking for information is a different transaction, and the answer tells you whether your list or your positioning is the thing that is broken.
The line most people get wrong: A third pitch wearing a question mark. If the answer could not change what you sell or who you send to, then it is not a question and the reader can tell.

How to put them in an order

A campaign is three of these, not one. Pick the opener that matches your best trigger, which for most B2B SaaS is the hiring email or the competitor switcher, then follow-up one three days later, then follow-up two six days after that, then the break-up. Four emails, twelve days, one thread. Then that person is done for ninety days whatever happens.

Run the same version at least fifty times before you change a word of it. On day three, after nineteen sends and no replies, you will want to rewrite the whole thing. Rewriting resets your sample to zero and teaches you nothing. Change one thing at a time: the list, then the subject, then the opener. Never two at once.

When a cold email campaign starts working, the hard part moves. It stops being the copy and becomes whether you sit down and send tomorrow, and the day after, and the fourteen days after that. That is where founders actually lose this, and it is why the first hundred customers playbook is mostly about showing up rather than tactics.

Questions founders actually ask

Do cold emails still work in 2026?: Yes, at lower volume and higher effort than in 2018. Inbox filtering got stricter, Google and Microsoft tightened bulk-sender rules, and buyers have seen every AI-written opener. What still works is a small list you chose deliberately, one researched line per email, and thirty to forty sends a day from a warmed domain. What stopped working is five thousand contacts and a spun first paragraph.
How many follow-ups should I send?: Two, then a break-up. That is four emails across twelve days, all in the same thread, each one adding something the previous email did not say. Sequences of seven and nine still exist and they still technically produce meetings, but the damage to your domain reputation and to the way people talk about you is real, and a founder cannot afford either.
What's a good cold email reply rate?: For a founder sending a hand-picked list in B2B SaaS, five to ten percent total replies is a working campaign and one to three percent positive replies is normal. Under two percent total, stop tuning the copy and change the list or the offer. Above fifteen percent, your list is probably warm rather than cold, which is good news you should learn from.
How long should a cold email be?: Between fifty and ninety words for a first touch, and under forty for a follow-up. The real test is whether the whole thing fits on a phone screen without scrolling, because that is where it will be read and where the decision to reply or archive is made.
Should I use AI to write cold emails?: Use it for the structure and the boring middle, never for the researched line. The opening sentence is the only part carrying proof that a person looked at their company, and a model that has not seen their job ads or their changelog cannot write it. Drafting with AI and researching by hand is the combination that holds up.
Is cold email legal?: In the US, CAN-SPAM permits B2B cold email if you identify yourself, give a real postal address, and honour opt-outs. In the UK and EU, business-to-business email to a corporate address is generally permitted under a legitimate-interest basis, with a clear identity and an easy opt-out, though rules differ by country and are stricter for sole traders and partnerships. Check your own jurisdiction before you scale, and always honour the first unsubscribe.