Four or five. Send 3 or 4 follow-ups after the first cold email, so 4 or 5 emails in total, with at least 3 days between them. Then stop.
Five emails is the only count that sits inside every range the big cold email platforms published in 2026, once you convert them to the same unit. lemlist's advice works out to 5 to 10 emails, Woodpecker's to 4 to 7, and Instantly's to 4 to 7 in its benchmark report and 4 to 5 on its blog.
The follow-ups are where a large share of the replies live. Across campaigns sent through Instantly in 2025, 42% of all replies came after the first email. Past the fifth email the gain per send gets small, while every extra message is one more chance for someone to press "report spam", and Gmail gives you very little room for that.
If LinkedIn runs next to your email, 4 emails is enough, because the other channel carries the rest of the touches.
Why do the recommended follow-up counts disagree?
They mostly disagree on the unit, not the number. One source counts follow-ups, another counts touchpoints, a third counts total attempts. Here are six published recommendations, converted to total emails including the first one:
| Source | What it counts | Its advice | Total emails | What the advice rests on |
|---|---|---|---|---|
| lemlist, updated August 2026 | Follow-ups | 4 to 9 | 5 to 10 | "Millions" of lemlist campaigns, no date range given |
| Woodpecker, updated June 2026 | Touchpoints | 4 to 7 | 4 to 7 | More than 20 million emails sent through its platform |
| Instantly benchmark report, January 2026 | Touchpoints | 4 to 7 | 4 to 7 | Thousands of workspaces, January 1 to December 18, 2025 |
| Instantly blog, February 2026 | Follow-ups | 3 to 4 | 4 to 5 | Its own 2025 analysis plus one agency's study of 16.5 million emails |
| One lead gen agency's template guide | Follow-ups | 3 to 5 | 4 to 6 | Guidance in a template article |
| Another agency's template guide | Total attempts | 5 to 7 | 5 to 7 | Guidance in a template article |
The first four rows come from platforms that sell the software the sequences are sent with, and three of them rest on their own users' campaigns. The data can still be right. Just read the upper ends of the ranges with some care.
Put the rows side by side and the spread shrinks. The low ends sit at 4 or 5 emails. The high ends run from 5 to 10. The only total that fits inside all six ranges is 5 emails.
How much does each follow-up add?
The lemlist article gives a reply rate for each position in a sequence, which makes the drop-off easy to see:
| Email in the sequence | Reply rate (lemlist data) |
|---|---|
| Email 1 | 4.5% |
| Email 2 | 3.75% |
| Email 3 | 2.75% |
| Email 4 | 2.1% |
| Emails 5 and later | Lower with each step |
| Sequence of up to 10 emails, total | Up to 22.37% |
Add up the first four step rates and you get 13.1%. The lemlist figure for a sequence running to 10 emails is 22.37%. So emails 5 to 10 together add about 9 more points, roughly 1.5 points each, against 4.5 for the first email.
On a list of 1,000 people, if the step rates add up the way lemlist's total implies, that is about 131 replies from four emails. The next six emails bring about 93 more, around 15 per send. Worth something, but each of those sends also carries the spam risk covered further down.
Instantly's 2025 numbers show why stopping after one email is expensive. The first step produced 58% of all replies in its data and the follow-ups produced the other 42%. Woodpecker's statistics page reports the same direction: campaigns with 3 to 5 follow-up steps reached an 8.3% reply rate, against 4.1% with no follow-up, though it does not say which dataset those figures come from.
One study, reported two ways
Be careful with one agency's analysis of 16.5 million emails. Woodpecker and Instantly both cite it, and both quote an 8.4% figure from it. Woodpecker says 8.4% was the reply rate of the first follow-up, often the best single step. Instantly says 8.4% was the reply rate of single-email campaigns.
Same study, same number, two different meanings, so treat that 8.4% as unreliable in both directions.
Where Woodpecker and Instantly do agree is the range: both name 4 to 7 touches in total as the target, Instantly in its benchmark report. Instantly's blog also gives 27% for 4 to 7 emails against 9% for 1 to 3, but that figure traces to Woodpecker's own 20-million-email data, not the 16.5-million-email study above.
Why does the third follow-up still matter when replies drop?
The per-step numbers fall with every email, so it is tempting to cut the sequence at two. That is where the data and what I see in real campaigns part ways.
Sending one message and waiting is a lottery ticket, and two messages are not enough either.
Smirnov Consulting Group is a Prague-based B2B outbound lead generation agency that runs cold email and LinkedIn campaigns for founder-led B2B companies and books qualified sales calls. In those campaigns, the meeting tends to land on the third follow-up, but only after real time has passed between messages.
By then, most teams have already marked the lead cold and moved on to fresh names.
Why would a later email beat an earlier, better-written one? Because time does work that copy cannot. Between your first email and your fourth, the prospect's quarter changes, a vendor lets them down, a problem lands on their desk. The fourth message arrives on a different day in a different mood.
I walk through the full build of that sequence, step by step, in how my team structures a follow-up sequence.
Two notes on reading the data against this. First, the vendor tables count replies, and a reply can be "not interested" as easily as "yes". Each dataset also counts replies its own way, which I compare in a breakdown of cold email reply rate benchmarks. What I care about is the booked meeting.
Second, count touches across channels. Budget for 5 or 6 as the floor, and spread them over two channels where you can. Cold email usually gets four steps in that mix, with LinkedIn carrying the rest. Follow-up statistics traced back to a real sample, across calls and email, land at roughly 5 to 8 touches.
When do more follow-ups start to hurt deliverability?
Every follow-up goes to the people who did not answer the last one. Some of them will mark it as spam, and that number is the real ceiling on sequence length.
Gmail's sender guidelines set 0.3% as the spam rate every sender must stay below, measured in Postmaster Tools. Their advice is to aim for less than 0.1% and to never touch 0.3%. That is 1 spam report per 1,000 emails as the target, and 3 per 1,000 as the line. Guideline checked September 2026.
The platforms say the same thing in their own words. The lemlist article says that after 9 or more follow-ups the benefit is negligible and you risk being marked as spam. Woodpecker's own guidance says spam and unsubscribe risk triples beyond four follow-ups.
Instantly warns that excessive follow-ups raise spam complaints, while spaced follow-ups that add something new do not hurt deliverability, in its view.
Put that next to the reply table. Emails 2 to 4 still add real replies. Emails 6 to 10 add about 1.5 points each while the complaint risk climbs. Five emails is a sensible place to stop, and your domain setup decides how much margin you have, which is why the boring part of sending infrastructure comes before any sequence.
A five-email schedule built from the published cadences
Spacing advice is close across the sources too. Instantly's report says 3 to 4 days between touches. Its blog says to wait 3 days before the first follow-up and reports that doing so lifts replies by 31% over next-day follow-ups, a figure it attributes to industry research without naming the dataset. The lemlist gaps are 2 days, then 4, then 4, then 5 or more.
Here is a schedule that starts with the 3-day gap and widens it the way lemlist does:
| Send on | Its job | A line you could adapt | |
|---|---|---|---|
| 1 | Day 1 | The offer and one small ask, such as a short call | "Worth 15 minutes next week to see if this fits?" |
| 2 | Day 4 | A different angle on the same problem | "Coming at this from a different side:" |
| 3 | Day 8 | Proof, such as one client result with a number | "One result from a company close to yours:" |
| 4 | Day 13 | One short question about a different pain | "Quick question: is winning new clients a priority this quarter, or later in the year?" |
| 5 | Day 19 | A clear last note that closes the loop | "I will leave it here. If this becomes a priority later, a one-word reply is enough." |
Each step needs its own reason to exist. If an email only says "just checking in", cut it. With LinkedIn running alongside, drop one email and let a connection request and a message take its place.
When should you stop following up?
Stop the sequence on any of these, the same day:
- Any reply at all. A yes goes to booking. A "not now" goes into your CRM with a date to come back, and leaves the sequence.
- A request to be removed. Remove the person from every list and every sequence.
- A bounce. Take the address out and check the rest of that list's source.
- The last planned step. When email 5 goes unanswered, the sequence is finished.
Other follow-up questions founders ask
Is sending four follow-ups annoying?
To some people, yes, and they will tell you or press "report spam". The data still favors following up: 42% of replies in Instantly's 2025 data came after the first email. Keep each email short, give it a new reason, and watch your spam rate against Gmail's 0.3% line.
How long should you wait between cold email follow-ups?
At least 3 days before the first follow-up, then let the gaps grow. The platforms land between 2 and 5 or more days per gap. The later follow-ups work best once real time has passed between messages, so do not squeeze five emails into one week.
What counts as a touchpoint?
Any separate contact attempt: a cold email, a LinkedIn connection request, a LinkedIn message. That is why touchpoint counts run higher than email counts. Five to six touches across email and LinkedIn can mean only four emails.