Most sales email advice fixates on the wrong thing. It obsesses over subject lines and clever openers while ignoring the two variables that actually move reply rates: whether the person you emailed has the problem you're describing, and whether you gave them a reason to answer today rather than never.
A good sales email is a research artefact more than a writing exercise. The copy matters, but it matters roughly a third as much as targeting and timing. Teams that fix targeting first routinely see reply rates go from under 2% to somewhere between 6% and 12% without changing a single word of their template. Teams that only rewrite copy tend to stay flat and conclude that outbound is dead.
This guide covers what actually works: how to build a list narrow enough to write to, how to structure the email itself, how many follow-ups to send and when, what to do after a call, and how to tell whether the whole programme is worth running.
Start with a list you could describe in one sentence
If you can't finish the sentence "everyone on this list is a company that just..." then the list is too broad and no amount of copywriting will save it.
The best sales email lists are built around a trigger — an observable event that suggests the problem is live right now. A few that consistently produce:
- Hiring signals. A company posting for a demand generation manager has an acknowledged gap. A company posting three SDR roles is scaling outbound and will need tooling.
- Tech stack changes. Adding or removing a tool in your category is the strongest buying signal available, because someone has already decided the status quo isn't working.
- Funding and expansion. New capital means new budget lines, usually within 90 days.
- Public pain. Job listings, review sites, changelogs, and conference talks all leak information about what a team is struggling with.
- Behavioural signals from your own site. Someone who read three pricing-adjacent pages last week is a different prospect from a cold name pulled off a database.
Fifty accounts built around a live trigger will beat five thousand pulled from a filter on employee count. This is unintuitive to teams measured on volume, but the arithmetic is straightforward: a 400-email campaign at 9% reply is 36 conversations, while a 5,000-email blast at 0.8% is 40 conversations that cost far more to produce, burn your domain reputation, and generate a pile of unsubscribes and spam complaints along the way.
Domain health is the quiet constraint here. Every low-relevance send teaches inbox providers that your mail is unwanted, and that reputation applies to your marketing sends too. Outbound and lifecycle email share the same infrastructure, which is why the segmentation discipline described in our guide to targeted email marketing matters just as much on the sales side.
The anatomy of a sales email that earns a reply
Cold email has roughly eight seconds of attention on a phone screen. The structure below is built for that constraint.
Subject line: descriptive, not clever. "Question about your Q4 SDR hires" outperforms "Quick one 👋" because it tells the recipient what the email concerns. Clever subject lines get opened and deleted. Descriptive ones get opened and read. Keep it under about 50 characters so it doesn't truncate on mobile.
First line: proof you did the work. This is the single highest-leverage sentence in the email, and it must not be about you. "Saw you're hiring two enterprise AEs in the northeast" works. "I hope this email finds you well" is a signal that a machine sent it. Never open with your own company name.
Second and third lines: the problem, stated in their language. Describe the specific failure mode your product prevents, using the vocabulary the buyer uses internally. If you sell attribution software, don't say "unlock full-funnel visibility" — say "you can't tell finance which of last quarter's channels produced the pipeline." Specificity is credibility.
Fourth line: evidence, quantified and comparable. One relevant result from a company that looks like theirs. A number with a denominator beats an adjective every time. "Cut cost per meeting from $340 to $190 across two quarters" is worth more than "dramatically improved efficiency."
Close: one small, specific ask. The mistake almost everyone makes is asking for a 30-minute demo from someone who has known you for eight seconds. Ask for something proportional to the relationship: "Worth a look?" or "Is this on your list for Q1, or handled?" An interest-check question is easier to answer than a calendar negotiation, and a "no, we sorted it" reply is genuinely useful because it clears your pipeline.
Total length: 75 to 125 words. Anything longer gets skimmed for the ask and discarded. Write it, then cut every sentence that would still be true if you sent it to a different company.
A few things to strip before sending: attachments, multiple links, images, tracking pixels on the first touch, and anything that renders as a marketing template. Plain text from a real person's mailbox lands in the primary tab far more often than an HTML layout that looks like a newsletter.
Personalisation that scales without lying
There's a real tension between personalisation and volume. The resolution isn't to fake personalisation with mail-merge variables — recipients spot {{first_name}}, I loved your post about {{topic}} instantly — it's to personalise at the segment level and vary only the top line.
Build ten to fifteen accounts per segment where the segment shares a genuine problem. Write one strong email for that segment. Then personalise only the opening sentence per account, using the trigger you built the list on. That gives you 90% of the lift of full personalisation at maybe 20% of the effort.
AI drafting tools are useful here, with one caveat: they're good at rewriting and summarising research, and bad at deciding who to email. Use them to compress a prospect's recent announcements into a usable opening line, not to invent an angle. We covered where these tools help and where they mislead in our review of AI tools for sales.
The follow-up sequence is where most replies actually come from
Analysis of large outbound datasets consistently shows the majority of replies arrive after the first email. A single-touch campaign leaves most of its return on the table.
A sequence that works for most B2B contexts:
- Day 0 — the initial email. Full structure as above.
- Day 3 — a new angle, not a bump. Do not send "just following up" or "bumping this to the top of your inbox." Send a different reason to care: a second use case, a relevant benchmark, a customer story from their vertical. Three sentences maximum.
- Day 8 — proof. A short case study, a specific number, or a one-line teardown of something on their site. This is the touch that most often converts a lurker.
- Day 15 — a lateral move. If the original contact hasn't replied, try one adjacent person in the same org. Never CC both.
- Day 22 — the close-out. "I'll assume the timing isn't right and stop here — happy to reconnect next quarter." Polite close-outs generate a surprising number of replies, because they remove the obligation and often surface a real "actually, ask me in March."
Then stop. Beyond five or six touches you are converting a small number of extra meetings into a large amount of brand damage, and each additional send raises your spam complaint rate. Re-approach the same account in 90 days with a genuinely new reason, not the same sequence restarted.
Keep every follow-up in the same thread with the same subject line. It gives the recipient context and it reads as one conversation rather than a campaign.
Follow-up after a sales call is a different job
Post-call follow-up is not outbound, and it should not read like it. Its job is to make the buyer's internal case for them, because the person you spoke to has to sell this to three colleagues who weren't on the call.
Send it within 24 hours, ideally the same afternoon. Include:
- What they told you their problem was, in their words. This confirms you listened and gives them language to reuse.
- What you agreed the next step is, with a date and an owner.
- One asset that answers the objection they raised, not a generic deck.
- Anything they can forward without edits. If your follow-up email can be pasted into a Slack channel and still make sense to someone who wasn't there, it will get forwarded.
Then confirm the next step exists in a calendar. A follow-up email that ends with "let me know what works" reliably produces nothing. One that ends with "I've sent an invite for Thursday at 2 — move it if that's awkward" produces a meeting.
Speed matters more than polish here, and it matters more than most teams believe. Response time to an inbound or post-call signal is one of the few variables where the effect size is enormous and the fix is operational rather than creative — a point we covered in detail in our guide to lead routing software.
Measuring whether sales email is actually working
Open rates have been unreliable since mail privacy protection started prefetching images, and they were always a weak proxy. Track these instead:
- Reply rate, split into positive, neutral, and negative. Negative replies are data, not failure — a high "not relevant" rate means your targeting is off, while a high "not now" rate means your timing is off. Those need different fixes.
- Meetings booked per 100 sends. The number that connects effort to pipeline.
- Meeting-to-opportunity rate. If you book plenty of meetings that never become opportunities, your messaging is attracting the wrong people, and the problem is upstream of the email.
- Bounce rate and spam complaints. Above 3% bounce, stop and clean the list. Above 0.1% complaints, stop entirely and fix targeting before you damage the sending domain.
- Cost per meeting and cost per opportunity. Rep time plus tooling plus data costs, divided by output.
That last one is where outbound programmes are usually won or lost, and it's worth running the numbers properly rather than eyeballing them. Add fully loaded rep cost to your data and sequencing tools, then work out what each closed customer costs you with a CAC calculator. Compare it against what those customers are worth over their lifetime using an LTV calculator — outbound tends to justify itself in markets with high contract values and long retention, and to fail quietly in low-ACV markets where the arithmetic never closes.
For sequence-level optimisation, a conversion rate calculator is useful for comparing variants honestly. Most teams declare a winner on 40 sends and a two-reply difference, which is noise. You need a few hundred sends per variant before a difference in reply rate means anything.
One structural point: sales email works best when it isn't the only thing touching the account. Prospects who've encountered your content before replying convert at materially higher rates than genuinely cold contacts, which is the practical argument for pairing outbound with the kind of asset programme described in our lead generation content marketing playbook. If you're weighing whether to build this capability in-house or buy it, our guide to choosing a B2B demand generation agency covers the diligence questions worth asking.
Compliance, briefly
In the US, CAN-SPAM requires accurate headers and sender information, a functioning opt-out, and a physical postal address. Cold B2B email is legal.
In the EU and UK, GDPR and PECR apply. B2B outreach can proceed under legitimate interest, but you need to document the assessment, honour objections immediately, and be able to explain where the data came from. Some member states are stricter than others. If you sell into Europe at any volume, get the position reviewed once rather than assuming US rules carry over.
Either way, honour opt-outs the same day and keep a suppression list that every tool in the stack reads from. Emailing someone who asked you to stop is the fastest way to turn a soft no into a complaint.
FAQ
How many sales emails should I send per day from one mailbox? Keep it under about 50 on a warmed domain, and ramp new mailboxes slowly over two to three weeks. Higher volumes need multiple domains and mailboxes, which adds cost and complexity. If your model only works at 500 sends a day per rep, the targeting is almost certainly too broad.
What's a good reply rate for cold sales email? Below 2% means targeting or messaging needs work. Between 3% and 5% is a functioning campaign. Above 8% usually means you've found a genuine trigger-based segment. Positive-reply rate is the number that matters, so track it separately from total replies.
Should I use a sales email template? Use a template for structure, never for content. The framework — trigger opening, specific problem, quantified proof, small ask — should stay constant. The specifics have to change per segment. A template sent unchanged to 200 companies reads exactly like what it is.
When should I stop following up? After five or six touches across roughly three weeks, close the loop politely and stop. Return in 90 days with a genuinely new reason to make contact. Persistence past that point costs you more in reputation and complaints than it produces in meetings.
Is cold email still effective in 2026? Yes, but the tolerance for volume-based outreach has collapsed. Inbox filtering is stricter, buyers are more cynical, and the gap between well-targeted and poorly-targeted outbound has widened enormously. The channel still works; sending a lot of generic email does not.