Cold email best practices
Six cold email best practices move the needle, roughly in this order of impact: protect your deliverability first (nothing else matters if you land in spam), write real examples instead of a generic template, earn the open with the subject line, personalize enough to prove you looked, follow up exactly once, and send at the right time. Get the first one wrong and the other five are wasted effort.
One of our clients, batch, ran these practices across its restaurant vertical: 7,550 emails sent, a 5.2% reply rate, 152 opportunities and $152,000 in pipeline.
Deliverability comes first
None of the other five practices matter if your emails never reach an inbox. Domain warmup, SPF/DKIM/DMARC records, and sending volume caps decide whether Gmail trusts you before a single prospect ever reads a word you wrote.
We rebuilt deliverability from zero for one of our clients, Bullseye for Schools, an EdTech company, after HubSpot's infrastructure tanked their domain reputation, restoring 100% inbox placement for over six months and producing 90 qualified leads. None of it was a copy change. As Greg puts it: deliverability is a system, not a tactic.
One thing that quietly wrecks placement: open tracking. Tracking opens means dropping a pixel in the email, and that pixel is exactly what triggers Gmail and Outlook's "images in this message are hidden - mark as spam?" banner before the reader's even read your first word. A practitioner who tested this directly - same copy, same list, same day, tracking on vs off - saw open rate look better with tracking on and reply rate come in roughly 5x worse (4.8% vs 0.8%). Judge a send time by replies, never by opens.
Open-tracking pixels trigger the "images are hidden" banner, and replies fall away with them.
We go deep on the technical setup, warmup schedules, and warning signs in .
Real examples beat templates
A copy-pasted template is recognizable within one sentence, and prospects have seen thousands of them. What actually gets replies looks nothing like the generic "I hope this finds you well" openers still circulating in 2026.
The customer-source opener, written for one of our clients, batch, a marketplace. Variant E hit a 10% reply rate - five opportunities from five sends in its test band.
Hey [Company]'s team,
I'm looking to partner with top [business type] in [City] that can host groups of 8 to 10 people looking for unique and fun experiences while visiting the city.
[Company] seems like an ideal fit.
We run a marketplace of over 3M people and would love to refer customers your way.
Would you be up for it?
What failed for the same client: a pain-led "here are two problems you have" framing with a lead magnet bolted on ran five times worse. Positioning as "we bring you customers" beat naming their problems.
The competitor-curiosity opener, written for one of our clients, Freshwater Creative, a creative agency. Beat two other variants by 2x in the same test. The 50 to 500 employee band produced 24 opportunities from 4,200 sends.
Hi [First name],
Noticed something interesting while looking at your competitor [Competitor]'s lead gen setup - a couple tactics that most [business type] in your market haven't picked up on yet.
Putting together a quick report showing the exact steps. Figured you'd want a look before everyone else catches on.
Want me to shoot it over?
The 500+ employee band hit a ceiling on the same copy: one opportunity from 6,382 sends. Company-size targeting mattered as much as the words.
Neither one says "I hope this finds you well", neither runs three paragraphs, and neither claims to help companies like yours. Both are under 70 words and end on a question that is easy to say no to.
We break down full real emails, line by line, in .
Subject lines that earn the open
The subject line has exactly one job: get opened. It doesn't need to sell anything, and trying to cram the pitch into it usually backfires - the ones that work most consistently read like a note from a colleague, not an ad.
The rule we write to is narrow on purpose: relevant to the email underneath it, all lowercase, three words or fewer, and it has to sound internal - like a line from a colleague inside their own company rather than a campaign.
That last part is the whole thing. Capitalised, clever, or longer than three words and it reads as marketing before the first word of the email gets a chance - the reader has decided what it is before they open it. Apply the rule to the two emails above and you get "group bookings" and "competitor setup": relevant, flat, and impossible to mistake for a campaign.
Full swipe examples and the patterns behind them live in .
Personalization that isn't fake
Merging in a first name isn't personalization, it's a mail-merge field. Real personalization proves you looked at the specific company: a recent hire, a product launch, a job posting that reveals what they're struggling with right now.
One relevant, current detail beats three paragraphs of researched flattery nobody asked for. Greg tested this directly once, 30,000 emails split between hand-researched paragraphs and three plain sentences that just nailed the right person and the right problem - the plain version won, and it scales in a way a researched paragraph never will. "The best cold email does not feel personalized. It feels like you read their mind."
So the effort goes into who you pick, not how much you write about them. As Adam puts it: you don't have a copywriting problem, you have a research problem.
"The results they've delivered have exceeded our expectations. We've been so impressed by their commitment to results that we're expanding their marketing efforts into our second company!"
The exact research shortcuts we use are in .
One follow-up, then move on
Most advice says the reply volume sits in follow-ups two through four. We don't run it that way. One follow-up, then the effort goes into contacting more people rather than chasing the same one. If someone hasn't answered twice, a third and fourth message in the same week won't change their mind, it just burns the domain you spent weeks warming. Anyone worth another go gets one about three months later, from a clean start.
The one we send asks for a referral rather than repeating the pitch:
Hey [First name], is there anyone else at [Company] I should be sending this to?
It does not guilt anyone for staying quiet, and it gives someone who is not the buyer an easy, useful thing to do. A wrong-person recipient can forward it in one line.
Adam puts the stopping rule best: getting ghosted is also a reply. It does not mean your prospect missed the email, or that they will come round on the fifth attempt. It means they could not be bothered. Move on, and do better next time. Persistence is a cadence, not a personality trait.
A good follow-up adds a new angle instead of just bumping the thread with "just following up." The exact cadence and message structure we run is in .
Timing: when to actually send
Timing is real but it's the smallest lever on this list - it can nudge reply rate a couple of points, not rescue a weak offer or a burned domain. Worth getting right once the first five practices are already solid, not before.
For what it is worth, the window we send in is mornings, 9am to 12, in the recipient's own time zone. The time zone half matters more than the hour.
Full send-time data by day and hour is in .
What are cold email best practices?
Six things in order of impact: deliverability first, then real examples over templates, subject lines that earn the open, honest personalization, a single follow-up, and the right send time.
What's the single biggest cold email mistake?
Ignoring deliverability. A brilliant email that lands in spam gets a 0% reply rate - it's the ceiling everything else operates under.
How personalized does a cold email need to be?
Enough to prove you looked at the specific company, not just a merged first name. One relevant detail beats a long researched paragraph.
How many follow-ups should a sequence have?
One, in our experience. Then move the effort to contacting more people rather than chasing the same one. If someone hasn't answered twice, a third and fourth message that week won't change their mind. Anyone worth another attempt gets one about three months later.
What reply rate counts as good?
1-3% is typical for a cold list. Our own client accounts run between 1.6% and 11%, and the spread comes down to targeting far more than wording.
See these practices run live.
We build the list and run a free 7-day campaign using everything on this page. No card, no contract.
Get my free meetingsP.S. If you can't remember the last time you checked your own domain's deliverability, that's probably where to start.