LetterDuck
Outreach

8 cold email templates that hold up

Three first-touch angles, two follow-ups, a break-up, a referral ask, and a re-engagement. Complete text, the variables that matter, and the one personalization rule that beats all the merge tags.

A cold email template is a starting structure, and the moment a thousand people send it verbatim it becomes a spam signature. That tension shapes everything below: eight complete templates, written by us, that you should retype in your own voice before a single one leaves your outbox.

Before any template: the infrastructure

The best cold email format in the world does not survive a domain the filters distrust. Since November 2025 Gmail rejects unauthenticated mail outright with a 550, and Microsoft has done the same since May 2025, so SPF, DKIM, and a published DMARC policy are the entry fee, not a nice-to-have. Our cold email infrastructure guide walks the whole setup, including why outreach belongs on a separate domain so an experiment cannot burn the domain your invoices ride on.

Volume discipline is the other half. The warm-up ramp we enforce in our own send path starts around 150 sends on day one and multiplies by roughly 1.4 per day to about 1,100 by day seven (Resend's documented domain ramp, codified into our throttle), then climbs 10 to 20% a day only while complaints stay under 0.1% and bounces under 2%. The send path auto-pauses itself at a 0.3% complaint rate because at that line Gmail withholds delivery mitigation until you are back under it for 7 consecutive days (Google bulk-sender rules, February 2024). A suppression list, checked before every single send, keeps bounced and complaining addresses from ever hearing from you twice.

One legal line, since templates travel: CAN-SPAM applies to commercial email at any volume. Accurate headers, a truthful subject, and an opt-out you actually honor, with penalties up to $53,088 per email under the FTC's January 2025 inflation adjustment. The details live in is cold email legal.

The personalization floor

Merge tags are not personalization. Every recipient of Hi {firstName}, I noticed {company} is doing great things in {industry} knows precisely what filled those braces, and the knowing is what kills the reply.

Our stance: one researched sentence beats five merge tags. A researched sentence is a fact about them that you found and they can check, like a broken link on their pricing page, a line from their changelog, or six SDR openings on their careers page. It cannot be scaled, which is exactly why it works: it proves a human spent minutes on them before asking for one back.

So each template below marks its research slot. The rest of the variables are mechanical (names, companies, dates). The research slot is the email. If you cannot fill it, you have a list, not a prospect.

Template 1: first touch, the observed problem

When to use: you found a real, checkable defect or gap in the prospect's world, and your product or service is the fix. The strongest possible opener because the first sentence is about them, provably.

Subject: {company}'s welcome email

{firstName}, I signed up for {company}'s trial on Tuesday. The welcome
email landed in my spam folder, and the day-two follow-up never
arrived at all.

I run {yourCompany}. We fix sending infrastructure for teams like
yours; this is usually a missing DKIM record or a burned shared IP,
and it takes under a week to correct.

Want the screenshots? Happy to send them over whether or not we ever
talk.

{yourName}

Variables: {firstName}, {company}, {yourCompany}, and the observation itself, which is the research slot. The observation must be true, dated, and specific enough that it could only be about them.

Why it works: it leads with their fact, not your pitch. The offer of free evidence ("want the screenshots?") gives them a reason to reply that costs nothing and admits nothing. Swap the deliverability specifics for whatever defect you actually found in your own niche.

Subject pairing: name the artifact, lowercase, no verb needed. Our cold email subject lines piece covers why the plain version outperforms the clever one.

Template 2: first touch, the routing question

When to use: you know the company fits but not whether this person owns the problem. Common early in a niche, or at companies whose org charts are opaque.

Subject: who owns deliverability at {company}?

{firstName}, quick one: when a campaign from {company} lands in spam,
whose desk does that land on?

Asking because {yourCompany} does one thing: we keep senders out of
the spam folder, and most teams we help did not have that person
until something broke.

If it is you, I will send a two-line audit you can run in five
minutes. If it is someone else, who should I ask?

{yourName}

Variables: {firstName}, {company}, {yourCompany}, plus your one-line category. Research slot: the question itself must name a problem this specific company plausibly has; a generic "who owns marketing?" earns a delete.

Why it works: a question aimed at their org itches until answered, and even "not me, ask Dana" is a reply. Replies matter beyond the conversation: mailbox providers weight replies above clicks and opens when scoring a sender, so a routing answer improves your domain's standing for the next send.

Subject pairing: the question verbatim. It fits under 40 characters with most company names and survives a phone screen.

Template 3: first touch, the trigger event

When to use: something changed at the company this month (a hiring wave, a launch, a migration, a deprecation deadline) and the change creates the problem you solve.

Subject: 6 SDR openings

{firstName}, your careers page lists six SDR openings. Six new
senders on one domain is the moment outreach starts hitting spam:
new accounts, no sending history, volume jumping week over week.

{yourCompany} sets up the domains, warm-up ramp, and daily caps
before a new team starts sending, so reply rates do not crater in
month two.

Is whoever runs sales ops thinking about this yet?

{yourName}

Variables: {firstName}, {yourCompany}, and the trigger, which is the research slot: their number, their announcement, their deadline, quoted accurately.

Why it works: the trigger supplies timing you did not have to invent. You are not arguing they have a problem; their own careers page is. The closing question routes to the right owner without demanding a meeting.

Subject pairing: the trigger's own number. 6 SDR openings is their fact in eleven characters, and it truncates nowhere.

Template 4: follow-up one, the added fact

When to use: three business days after the first touch, no reply. Sent as a reply in the same thread, keeping the original subject. Our cold email follow-up article covers the full sequence and spacing.

Subject: (reply in the same thread)

{firstName}, one thing I left out: {a fact or resource that stands
on its own, e.g. "the audit checklist we use is public, link below,
no signup"}.

It is useful whether or not we ever talk.

No pressure on the earlier note. If the timing is wrong, "not now"
is a perfectly good answer.

{yourName}

Variables: the added fact is the research slot, and the bar is strict: it must be worth reading on its own. "Just bumping this" fails the bar and trains the recipient to ignore you.

Why it works: each touch deposits value instead of withdrawing attention. Explicitly licensing a "not now" lowers the social cost of replying, and short negative replies still beat silence, for the conversation and for your sender reputation.

Subject pairing: none. Threading is the point; a new subject resets the context you spent touch one building.

Template 5: follow-up two, the smaller ask

When to use: five business days after follow-up one, and this is the last message in the sequence. The original framing may simply be wrong for them, so shrink the claim instead of repeating it louder.

Subject: (same thread)

{firstName}, different angle on my last two notes.

You may not care about {original framing} at all, and that is fair.
The one thing I would still check at {company} is {smaller specific
item}: it takes about ten minutes and costs nothing.

If those ten minutes turn up nothing, then I am wrong about the rest
too, and you should ignore me.

{yourName}

Variables: {original framing}, {company}, and the smaller item, the research slot: a genuinely quick, genuinely free check relevant to them.

Why it works: it converts a sales thread into a falsifiable claim. Offering the condition under which you are wrong is rare enough in cold mail to read as honest, because it is.

Subject pairing: same thread, no new subject. Three emails, one thread, one subject: that is the whole sequence's footprint in their inbox.

Template 6: the break-up

When to use: as the third and final touch, in place of template 5 when there is no smaller ask worth making. Roughly two weeks after first contact. One send, then actually stop; a break-up email you follow with more email is a lie.

Subject: (same thread)

{firstName}, closing the loop. Two notes without a reply usually
means no, not now, or wrong person, and every one of those is fine.

I will stop here. If {problem} surfaces later, this thread has
everything you need, and my door stays open.

{yourName}

Variables: {firstName}, {problem}. No research slot: this one earns its keep by ending cleanly.

Why it works: it returns control. Some recipients reply precisely because the pressure is gone; most will not, and the polite exit protects the asset that outlives any single sequence, your domain's reputation with that company's mail server.

Subject pairing: same thread, to the end.

Template 7: the referral ask

When to use: instead of template 2 when you suspect you have the wrong person and want a name, or after a "not me" reply. Never fabricate the mutual context; a fake referral is a lie with a witness.

Subject: wrong person for {topic}?

{firstName}, keeping this short because I may be aiming at the wrong
desk.

{One researched sentence: the company fact that made you write, and
the problem you solve in plain words.}

If this is yours, great, happy to share what I found. If not, would
you point me at the right person? One name is plenty, and I will not
claim you endorsed anything.

{yourName}

Variables: {firstName}, {topic}, and the researched sentence, the research slot.

Why it works: "I will not claim you endorsed anything" removes the exact fear that stops people from forwarding cold mail. You are asking for directions, not a favor, and directions are cheap to give.

Subject pairing: wrong person for {topic}? is disarming and true. Honesty as a subject strategy has a low bar to clear in this genre.

Template 8: the re-engagement

When to use: a lead who once replied and went quiet, or a closed-lost from 90 or more days ago. Something must have changed, on their side or yours, or you have no email.

Subject: (revive the old thread)

{firstName}, we talked in {month} about {topic} and you asked me to
check back when {condition}. That happened: {what changed, in one
line}.

Same offer as before with current numbers: {one concrete line}.

Still relevant, or has {company} solved this since?

{yourName}

Variables: {month}, {topic}, {condition}, and what changed, the research slot. Quote their words from the old thread if you can; nothing signals a real relationship like their own sentence returned to them.

Why it works: it resumes a conversation instead of starting a pitch, and the closing question makes "we solved it" an easy, useful answer. One hygiene note from our own playbook: if the address hard-bounced in the gap, it belongs on your suppression list, not in this sequence. We suppress hard bounces permanently and retry soft bounces about three times over 14 days, because a bounce rate above 2% is its own deliverability problem.

The anti-template warning

Every template in this article degrades with use, including ours. The mechanism has two halves. Recipients recognize the skeleton once the same shape has arrived from three different senders, and recognition converts curiosity into a spam report. Mailbox providers see the same body shape across thousands of senders, and matching a known mass pattern costs you the benefit of the doubt your small volume would otherwise enjoy. The complaint math from the top of this page does the rest: 0.3% is three reports per thousand sends.

So treat these as wireframes. Retype them in the words you would say aloud, reorder the parts, cut the lines that do not sound like you. Keep the thinking: lead with their fact, one research slot per email, one thread per sequence, a clean exit. What is a cold email, at its best, if not a short letter one person wrote to another? Our what is cold email piece makes that case in full.

Start with template 1, rewrite it in your own voice, and send it to ten well-researched people this week, inside the daily caps in how many cold emails per day. Read every reply before you send the next ten.