How Many Follow-Ups Should a Cold Email Sequence Have?

Most published guidance puts the right number at three to six follow-ups after the first email. Instantly's 2026 report finds 42% of replies come after step one, so follow-ups clearly pay. The ceiling is set by complaints, not replies: Gmail asks every sender to keep its spam rate below 0.3%, and each extra step is another chance to be reported.

There is no study that gives one correct number for every list. What there is: a few published datasets on where replies come from, a hard complaint threshold from Gmail and Yahoo, and some arithmetic you can do on your own sequence.

What the published data says about follow-ups

Every dataset we found agrees that follow-ups add a large share of replies, and that each later step adds less than the one before. The figures below are quoted from pages fetched on 1 October 2026.

Source	Figure	Caveat
Instantly, Cold Email Benchmark Report 2026	"58% of all replies are generated from step one"; follow-ups "contribute 42% of total replies"; "The sweet spot for sequence length is 4-7 touchpoints"	Platform-wide 2025 data; does not say how auto-replies are treated
lemlist	Reply rate 4.5% on email 1, 3.75% on email 2, 2.75% on email 3, 2.1% on email 4; 22.37% across 10 emails	No sample size, period or reply definition on the page; steps 5 to 10 not listed individually
Backlinko and Pitchbox	"A single follow-up message can boost replies by 65.8%"; "sending 3 or more messages results in the best overall response rate"	Link-building and PR outreach, not sales; page last updated April 2019

Two pieces of arithmetic make these more useful.

Instantly's split implies follow-ups add about 72% on top of the first email. If step one produces 58 replies in every 100, the follow-ups produce the other 42, and 42 divided by 58 is 0.72. That is in the same range as Backlinko's 65.8% for a single follow-up, from a different population and a different decade, which is mild reassurance that the effect is real.

lemlist's per-step figures shrink quickly. Email 4 gets less than half the reply rate of email 1 (2.1% against 4.5%). If those rates apply to the people still in the sequence, the first four emails together reach about 12.5% (1 minus 0.955 × 0.9625 × 0.9725 × 0.979), not the 13.1% you get by adding them up. The page does not say which way the figures were calculated, and it does not publish steps 5 to 10, so we could not reproduce the 22.37% figure for ten emails.

What none of these sources publishes is the other side of the ledger: spam complaints and removal requests by step. That is the number that decides where to stop. lemlist says that "after 9 or more follow-ups, the benefits are negligible, and you could be marked as spam", but gives no data for it.

The limit is complaints, not replies

Gmail's sender guidelines set a spam rate ceiling that applies to every sender, and a cold sequence is where you are most likely to hit it. Under "Requirements for all senders", Google's guidelines say: "Keep spam rates reported in Postmaster Tools below 0.3%." The Postmaster section is stricter: "Keep spam rates reported in Postmaster Tools below 0.10% and avoid ever reaching a spam rate of 0.30% or higher." Yahoo's sender best practices also say "Keep your spam rate below 0.3%".

The same Google page is plain about cold outreach in general: "Don't send messages to people who didn't sign up to get messages from you. These recipients might mark your messages as spam." Cold email exists in tension with that sentence. A sequence that keeps writing to people who never replied is leaning on it harder with every step.

Three details of how Gmail measures this matter for follow-ups:

1. It is daily. Google's sender FAQ states: "Spam rate is calculated daily." A sequence that fires steps 3, 4 and 5 for different cohorts on the same day adds them all into one day's rate.
2. The denominator is inbox deliveries, not sends. Postmaster Tools help defines spam rate as "the percent of your messages that are delivered to engaged recipient's Inbox and then marked as spam by the recipient." It also warns that if Gmail is already filtering much of your mail to spam, "the rate shown in the dashboard might seem low". A low number is not always good news.
3. Bulk senders lose help above 0.3%. The FAQ says bulk senders with a spam rate above 0.3% "will be ineligible for mitigation" until their rate stays below 0.3% "for 7 consecutive days".

What 0.3% means at cold email volumes

At typical per-inbox volumes, a single complaint can put a day over the line. This is plain arithmetic on Google's threshold:

Messages delivered to Gmail inboxes that day	Complaints that reach 0.1%	Complaints that reach 0.3%
100	0.1	0.3
300	0.3	0.9
1,000	1	3
3,000	3	9

At 300 delivered messages a day, one person pressing "report spam" is 0.33%. There is no such thing as an acceptable small number of complaints at that scale. This is why the per-day sending caps in how many cold emails per inbox per day and the step count of your sequence are the same decision.

How to decide where your sequence ends

Stop adding steps when the next one produces about as many complaints and removal requests as genuine replies. You can only find that point in your own data, because it depends on your list, offer and copy.

Here is a hypothetical example to show the method. The numbers are invented.

Step	Prospects reached	Human replies	Removal requests	Spam complaints (Postmaster or feedback)
1	970	30	3	0
2	940	15	4	0
3	925	8	5	1
4	917	5	6	1
5	912	3	7	2

In this example, step 4 is where removal requests overtake replies, and step 5 produces more complaints than interested people. A team looking at this would cut the sequence to three or four steps, rewrite step 4, or both. A team looking only at reply rate per step would see step 5 still producing replies and keep it.

To get these columns from a real campaign:

1. Export per-prospect results with the step at which each person replied, unsubscribed or stopped.
2. Classify replies by hand: interested, referral, not now, objection, removal request, auto-reply. Our post on reply rate benchmarks explains why auto-replies should not count.
3. Pull the daily spam rate from Google Postmaster Tools for the sending domain and line it up with the days each step went out. See our guide to reading Postmaster Tools.
4. Compare replies against removal requests plus complaints, per step.

Spacing between follow-ups

Published recommendations cluster around three to five days, but none of the sources we read publishes a controlled test of the gaps. Instantly says "Space touches 3-4 days apart". lemlist suggests a 2-day delay before email 2, then 4 days, 4 days, and "5+ days between follow-ups after the 4th email".

The reasoning for widening gaps as the sequence goes on holds up regardless of the exact numbers:

• Early gaps can be short because the first follow-up catches people who meant to reply and forgot. The message is still fresh.
• Later gaps should be longer because the people left have now ignored several messages. Writing to them every three days for a month is exactly the pattern that prompts a spam report.
• Spread cohorts across days so that your daily complaint denominator is not dominated by late steps. Because Gmail calculates spam rate daily, a day that is mostly step-5 sends is the day you are most likely to cross 0.3%.

A reasonable starting shape, offered as reasoning rather than evidence:

Step	Day	Purpose
1	0	The actual request, short and specific
2	+2 to 3	Same thread, one line, a nudge
3	+4 to 5	A new angle or a useful piece of information
4	+6 to 7	A clear close: say this is the last message

That is four emails in roughly two to three weeks. Lengthen it only when your own step data shows later steps earn their complaints.

When the sequence must stop

Any human reply, any unsubscribe request and any bounce should end the sequence for that person immediately. These are the stop conditions to configure in your sending tool:

• Any human reply, positive or negative. A "no thanks" followed by an automated follow-up is very likely to be reported as spam.
• Any removal or unsubscribe request, however phrased. Yahoo asks bulk senders to "Honor unsubscribes within 2 days"; Google recommends fulfilling them "within 48 hours". For cold email in the US, CAN-SPAM has its own opt-out rules; see CAN-SPAM and cold email.
• A hard bounce. The address does not exist; continuing hurts reputation for nothing. A 550 5.1.1 is the typical example.
• A reply from someone else at the same company, such as a referral or "please stop contacting our team". Stop at the domain level, not just the address.
• An out-of-office reply: pause and resume after the return date rather than stopping or carrying on.

Before the first step even goes out

Follow-ups multiply whatever is wrong with the list. An invalid address bounces on step 1 and, if your tool does not stop on bounces, on steps 2 to 5 as well. A catch-all domain may accept all five and bounce them later. Verifying before step 1 removes those rows from every step at once.

If a sequence was paused for months and you are restarting it, re-verify first, because addresses that were live when you built the list may not be now.

What to take from this

Follow-ups work: the published data consistently shows a large share of replies arriving after the first message. But the evidence for long sequences comes from vendors measuring replies, not complaints. Start with three or four follow-ups, measure removal requests and Postmaster spam rate by step, and cut the sequence at the step where complaints start to rival replies.

You can check a sample of your list for dead addresses with the free email verifier before step one goes out.