Most lists of cold email templates b2b teams find online are collections of copy you paste, send and forget. They optimise for open rate, which is the vanity metric of outbound. An open costs you nothing and earns you nothing. A reply is the only outcome that moves a deal forward.
This guide takes a different angle. Below are seven cold email structures used in real B2B prospecting, each with full copy, three subject line variations, a line-by-line explanation of why each sentence exists, and the follow-up cadence that belongs with it. If you understand the mechanics, you can rebuild any of these for your own market instead of pasting someone else’s words into your sequencer.
Why reply rate beats open rate in B2B outreach
Open tracking has become unreliable. Privacy proxies, image blocking and inbox pre-fetching inflate open numbers, so a 68% open rate can hide a 0.4% reply rate. Reply rate is the honest signal because a human had to make a decision.
Here is what a healthy B2B cold campaign looks like when you measure the right things:
| Metric | Weak | Acceptable | Strong |
|---|---|---|---|
| Reply rate (all replies) | Under 2% | 4% to 7% | 8% and above |
| Positive reply rate | Under 0.5% | 1% to 2% | 3% and above |
| Meetings per 1,000 sends | 0 to 2 | 5 to 10 | 15 and above |
| Bounce rate | Over 5% | 2% to 3% | Under 1% |
These are directional benchmarks for outbound to cold, verified, ICP-matched lists. Broad, unsegmented blasts sit far below the left column.
The anatomy every good cold email shares
Before the templates, the five components. Every structure below is a different arrangement of these same parts:
- The relevance trigger. One line that proves you are writing to this person and not to a list. It answers the silent question “why me, why now?”
- The problem framing. A specific, observable pain stated in the prospect’s vocabulary, not in your product’s vocabulary.
- The proof point. A named peer, a number, or a mechanism. Vague credibility (“trusted by hundreds of companies”) is invisible.
- The low-friction ask. The cheapest possible next step. “Worth a look?” costs less than “do you have 30 minutes Thursday?”
- The exit. Permission to say no. Counter-intuitively this raises reply rate because it removes the social cost of answering.
Length target: 50 to 125 words. Reading level: grade 5 to 7. One idea per email. One ask per email.
Personalisation tiers (choose deliberately)
| Tier | What it uses | Volume per rep per day | Best for |
|---|---|---|---|
| Tier 1: segment | Industry, role, company size, tech stack | 150 to 300 | SMB, high-volume, low ACV |
| Tier 2: signal | Hiring posts, funding, product launches, job changes | 40 to 80 | Mid-market |
| Tier 3: research | Podcast quotes, teardowns, account-specific analysis | 10 to 20 | Enterprise, named accounts |
The 7 cold email templates for B2B outreach
1. The trigger event email
Use when: a public signal gives you a legitimate reason to write today (funding round, new hire in a relevant function, office opening, tooling change, regulatory deadline).
Subject line variations
- question about the {{role}} hire
- {{Company}} + {{trigger}}
- saw the {{trigger}} news
Full copy
Hi {{FirstName}},
Saw you opened two SDR roles in Manchester last week. Usually that means the pipeline target moved before the process was ready for it. The piece 10 Best B2B Cold Sales Email Templates makes a good next read.
When {{PeerCompany}} scaled from 3 to 9 reps, their ramp time was the bottleneck: new hires took 11 weeks to hit quota. We got that to 6 by rebuilding the first-30-days call library.
Is ramp speed on your list this quarter, or is it further down?
{{SenderName}}
Why each line works
- Line 1 (the trigger): specific, verifiable, recent. It answers “why now” in eight words and proves a human looked at the account.
- Line 2 (the inference): you do not just report the signal, you interpret it. The interpretation is what makes you sound like a peer rather than a scraper.
- Line 3 (the proof): a named comparable company plus two numbers. Numbers with a before and after are far stickier than a single claim.
- Line 4 (the ask): a priority question, not a calendar question. Answering “further down” is still a reply and still qualifies the account.
Follow-up cadence
| Step | Day | Content |
|---|---|---|
| 1 | 0 | Trigger email above |
| 2 | 3 | Same thread: the ramp checklist as a plain-text list, no attachment |
| 3 | 8 | New thread: second trigger or a different angle on the same pain |
| 4 | 15 | Permission close (see template 7) |
2. The problem, agitate, solve (PAS) email
Use when: your ICP shares one loud, well-understood operational pain and you have no account-specific signal.
Subject line variations
- {{painful metric}} at {{Company}}?
- the invoice chasing problem
- quick one on renewals
Full copy
Hi {{FirstName}},
Most finance teams in {{industry}} still chase overdue invoices by hand through a shared inbox.
It works until volume doubles. Then DSO drifts out, the same three customers get chased twice, and someone spends Friday afternoon reconciling what was actually paid.
We automate the chasing sequence and reconcile against the ledger, so the inbox is only for exceptions. {{PeerCompany}} cut DSO by 9 days in one quarter.
Is manual chasing still how it works on your side?
{{SenderName}}
Why each line works
- Problem: stated as an observation about the category, not an accusation about them. Nobody defends against a category statement.
- Agitate: three concrete consequences, each one a scene the reader can picture. Agitation fails when it is abstract (“this costs you money”) and works when it is cinematic (“Friday afternoon reconciling”).
- Solve: one sentence on mechanism, one sentence on result. Mechanism first is important: buyers believe outcomes only when they can see how the outcome was produced.
- Ask: a yes/no diagnostic. It is easy to answer and it segments your list for you.
Follow-up cadence
- Day 0: PAS email.
- Day 4: same thread, one-line nudge with a second consequence you did not use.
- Day 9: new thread, swap the pain for the adjacent pain (if invoice chasing did not land, try cash forecasting).
- Day 16: permission close.
3. The peer proof email (case study, done properly)
Use when: you have a customer that looks unmistakably like the prospect and you can share a specific number.
Subject line variations
- {{PeerCompany}} {{result}}
- how {{PeerCompany}} handled {{problem}}
- similar setup to {{PeerCompany}}?
Full copy
Hi {{FirstName}},
{{PeerCompany}} runs the same setup you do: {{shared attribute, e.g. two ERPs after an acquisition, 40-person field team}}.
Their ops lead was spending roughly a day a week reconciling stock between systems. We connected the two, and that day went back to the team. Error rate on picks dropped from 4% to under 1%.
If reconciliation is eating similar time at {{Company}}, I can send the two-page write-up of how they did it. No call needed.
{{SenderName}}
Why each line works
- The mirror: naming the shared attribute is the personalisation. It tells the reader the case study is relevant before they decide whether to care.
- The human detail: “a day a week” beats “efficiency gains” because it converts into the reader’s own calendar instantly.
- Two numbers, one soft one hard: time saved is relatable, error rate is defensible in front of their boss.
- “No call needed”: this is the single highest-leverage phrase in the template. It removes the assumption that replying means surrendering 30 minutes, and it converts the ask into a content request that a busy person can grant in three seconds.
Follow-up cadence
Because the ask is content, the follow-ups deliver, they do not chase.
| Step | Day | Content |
|---|---|---|
| 1 | 0 | Peer proof email |
| 2 | 3 | Send the summary anyway, in the body, unrequested |
| 3 | 7 | The one thing that nearly derailed the peer’s rollout (candour builds trust) |
| 4 | 14 | Permission close |
4. The one-question email
Use when: you are emailing senior people, or your previous sequences are being ignored. This is the shortest structure and often the highest reply rate per word.
Subject line variations
- {{FirstName}}, quick question
- {{specific process}} at {{Company}}
- who owns this?
Full copy
{{FirstName}}, who owns supplier onboarding at {{Company}} these days? saleshandy.com has a solid rundown on this.
Asking because we shortened it from 6 weeks to 11 days for {{PeerCompany}} and I do not want to send this to the wrong person.
{{SenderName}}
Why each line works
- Under 45 words: it reads fully in the preview pane, so the open and the read are the same event.
- A routing question, not a selling question: senior people answer routing questions reflexively because forwarding is easier than declining.
- The justification clause: “asking because” is a compliance trigger. A request with a stated reason outperforms the same request without one.
- The number arrives sideways: the value proposition is smuggled inside the explanation, so it never feels like a pitch.
Warning: do not use this if you already know the person owns the area. Fake ignorance is transparent and it burns the account.
Follow-up cadence
- Day 0: the question.
- Day 3: “Should I be talking to {{Name}} in ops instead?” (naming a real colleague you found on LinkedIn).
- Day 9: the full pitch, now that you have earned the right to make one.
- Day 17: permission close.
5. The value-first teardown email
Use when: you sell into a function whose output is public (website, careers page, app store listing, ad library, packaging, SEO footprint) and you can produce a genuine observation in under ten minutes.
Subject line variations
- 3 notes on your {{asset}}
- {{Company}} {{asset}}: one thing I would change
- found this on your checkout
Full copy
Hi {{FirstName}},
I went through your checkout on mobile this morning. Three things stood out:
1. The address form asks for county before postcode, which breaks autofill on iOS.
2. The discount field is above the payment button, so shoppers leave to hunt for codes.
3. No express wallet option, which is usually worth 4 to 7 points of mobile conversion.Number 2 is the cheap one. Want me to send the 90-second Loom showing where it happens?
{{SenderName}}
Why each line works
- Proof of effort: “this morning” plus device-specific detail is impossible to fake at scale, which is precisely why it converts.
- Three findings, not one: one finding looks like a fluke, five looks like a bulk audit. Three reads as considered.
- You give away the fixes: the reciprocity is real only if the advice is usable without you. Withholding the insight to force a call destroys the whole mechanism.
- The ask is a Loom, not a meeting: asynchronous, zero calendar cost, and it moves the relationship into a channel where you can demonstrate expertise.
This template does not scale past roughly 15 emails a day per person. That is a feature. Reserve it for your top-tier accounts.
Follow-up cadence
- Day 0: the teardown.
- Day 2: send the Loom without waiting for permission. Effort already spent should not be held hostage.
- Day 8: a fourth finding, or the same finding observed at a competitor.
- Day 16: permission close.
6. The referral and internal redirect email
Use when: you have entered the account at the wrong level, or you want to reach a director through a VP. Also the best structure for turning a warm connection into a cold-but-credible open.
Subject line variations
- {{ReferrerName}} suggested I reach out
- passed along by {{ReferrerName}}
- {{FirstName}} + {{ReferrerName}}
Full copy
Hi {{FirstName}},
{{ReferrerName}} mentioned you took over the reporting stack after the reorg.
We built the attribution layer that {{ReferrerName}}’s team uses to close the loop between paid spend and pipeline. It took two weeks and no data engineering time on their side.
Happy to show you what they see, or to stay out of your way if this is already solved. Which is it?
{{SenderName}}
Why each line works
- Borrowed trust in the first six words: the referrer’s name does the work that ten personalisation tokens cannot.
- Context on the transition: “after the reorg” shows the referral was a real conversation, not a name you scraped.
- Cost removal, not benefit stacking: “no data engineering time” pre-empts the actual objection of a technical buyer, which is effort, not price.
- The binary close: offering “stay out of your way” as a legitimate option is what makes the other option feel chosen rather than pushed.
Rule: never invent or stretch a referral. If the referrer would not recognise the sentence, do not write it.
7. The permission close (the breakup that is not rude)
Use when: it is the final step of any sequence. In most campaigns this single email produces 20% to 35% of total replies.
Subject line variations
- closing the loop
- should I stop?
- last one from me
Full copy
Hi {{FirstName}},
I have sent a few notes about {{problem}} and heard nothing back, which usually means one of three things: it is not a priority, someone else owns it, or the timing is wrong.
A one-word reply tells me which and I will act accordingly: not a priority, wrong person, or later.
Either way, this is the last one from me.
{{SenderName}}
Why each line works
- Naming the silence: acknowledging the non-response defuses the awkwardness that keeps people from answering at all.
- Three options instead of an open question: multiple choice reduces the cognitive cost of replying to near zero.
- A one-word reply is explicitly allowed: people who felt they owed you a paragraph now owe you a word.
- A real ending: the promise to stop must be kept. If you re-sequence the same person two weeks later, the technique stops working across your whole list.
Matching template to situation
| Template | Personalisation tier | Best fit | Typical ask |
|---|---|---|---|
| 1. Trigger event | Tier 2 | Mid-market, fast-moving accounts | Priority question |
| 2. PAS | Tier 1 | High-volume SMB | Yes/no diagnostic |
| 3. Peer proof | Tier 1 or 2 | Crowded categories | Send the write-up? |
| 4. One question | Tier 1 | C-level, unclear org charts | Routing |
| 5. Teardown | Tier 3 | Named enterprise accounts | Async video |
| 6. Referral | Tier 3 | Multi-threading, expansion | Binary choice |
| 7. Permission close | Any | Final step of every sequence | One-word reply |
Follow-up rules that apply to all seven
- Four to five steps over 14 to 18 days. Beyond that, incremental replies collapse and complaint risk rises.
- Change the angle, not the volume. “Just bumping this to the top of your inbox” is a wasted send. Each follow-up should introduce one new idea.
- Alternate threads. Steps 1 and 2 in the same thread, step 3 in a new thread with a new subject, step 4 in whichever thread got more engagement.
- Send on the recipient’s clock. Tuesday to Thursday, 7:30 to 9:30 local time, still outperforms mid-afternoon for most B2B roles.
- Never attach anything cold. Attachments and heavy tracking hurt deliverability. Paste the content in the body.
Deliverability: the part that invalidates every template if you skip it
The best copy in the world cannot outrun the spam folder. Before you launch:
- Send from a separate domain (for example getcompany.com, not company.com) so your main domain reputation is insulated.
- Configure SPF, DKIM and DMARC. Bulk senders to major providers are now expected to have all three, plus a one-click unsubscribe on marketing-class mail.
- Warm each mailbox for two to three weeks and cap it at 20 to 30 cold sends per day. Multiple mailboxes, not bigger mailboxes.
- Verify every address before sending. Keep bounces under 2%.
- Plain text only. No images, no tracking pixels, one link at most, and only after the first email.
- Monitor spam complaints under 0.1%. Above 0.3% you will be filtered regardless of copy quality.
Seven mistakes that kill reply rates
- Opening with yourself. “My name is X and I work at Y” spends the most valuable line in the email on information the reader can get from your signature.
- Fake personalisation. “Loved your recent post!” without naming the post is worse than no personalisation, because it signals automation.
- Two asks in one email. A call and a resource and a LinkedIn connect gives the reader a decision to make instead of an action to take.
- Feature lists. Buyers do not buy capabilities, they buy the removal of a specific friction they already feel.
- Calendar links in email one. Booking implies a decision you have not earned yet.
- Praising the prospect too hard. Flattery reads as a prelude to a pitch, and it is.
- Sending to a list you did not qualify. Copy cannot fix targeting. If reply rates are under 2%, look at the list before you rewrite the email.
How to test these templates properly
Test one variable at a time, on a big enough sample to mean something:
- Sample size: minimum 200 contacts per variant, ideally 400.
- Variable order: list segment first, then the opening line, then the ask, then the subject line. Subject line testing is the last lever, not the first.
- Measure positive replies, not total replies. “Not interested” is data, but it is not progress.
- Give it 14 days before judging, so the whole cadence has run.
- Retire winners. A structure that beats everything for six months will get copied by your competitors and fatigue your market. Rotate before performance decays.
FAQ
How long should a B2B cold email be?
Between 50 and 125 words for the first touch. The email should be readable in the preview pane without scrolling on a phone. Longer emails can work for the teardown structure, but only because the extra length is proof of work rather than extra pitch. See leadfeeder.com for their take.
Should I personalise every cold email individually?
No. Match the personalisation tier to the account value. Tier 1 segment personalisation is enough for SMB volume plays. Reserve Tier 3 individual research for accounts where a single closed deal justifies an hour of prep.
How many follow-ups should a cold email sequence have?
Four to five total touches spread over 14 to 18 days, ending with the permission close. Reply rates from step six onwards are usually not worth the deliverability and reputation cost.
What is a good reply rate for B2B cold email?
4% to 7% total replies is acceptable on a well-targeted list, 8% and above is strong. Positive reply rate matters more: 1% to 2% is workable, 3% and above is excellent.
Are cold emails legal for B2B?
In the US, CAN-SPAM allows unsolicited commercial email provided you identify yourself, use accurate headers, include a physical address and honour opt-outs. In the EU and UK, B2B outreach is generally permitted under legitimate interest but you must be relevant, identify the source of the data on request, and provide a clear opt-out. Check your own jurisdiction and never email addresses obtained from breached or scraped consumer sources.
Do subject lines really matter if I optimise for replies?
They matter for getting the email read, not for getting it answered. Keep them lowercase, two to five words, specific, and free of anything that reads like marketing. The body wins the reply.
Should I use AI to write cold emails?
Use it to research accounts, summarise signals and draft variations, then rewrite the opening line and the ask yourself. Fully generated emails converge on the same phrasing everyone else is sending, which is exactly what your prospect’s inbox is learning to ignore.
Putting it to work
Pick two of these seven structures, not all seven. Choose the one that matches your best available signal and pair it with the permission close. Build a list of 200 genuinely qualified contacts, run the cadence for two weeks, and measure positive replies only. Then change one thing.
The templates are the easy part. The reply comes from the relevance of the first line, the specificity of the proof, and the smallness of the ask. Get those three right and almost any structure works.