How many follow-up emails should you send?
Two, for most small teams: one first email and two follow-ups, then a pause. That is our recommendation, and it sits inside the range the 2025 and 2026 data supports. The sources that publish numbers recommend anything from two emails to seven, and each looked at a different group of senders:
| Source | What it analysed | What it recommends |
|---|---|---|
| Instantly, Cold Email Benchmark Report 2026 (software vendor, updated 12 January 2026) | Billions of cold email interactions across thousands of workspaces, by Instantly's description, sent 1 January to 18 December 2025 | 4 to 7 touchpoints |
| Belkins, sales follow-up statistics (outbound agency, updated 26 June 2026) | 7,530,489 emails from its own client campaigns in 2025 | 3 to 5 steps |
| Smartlead, State of Cold Email 2026 (software vendor, no date shown, checked 20 October 2026) | More than 850 million emails from senders with at least 500 sends; period not stated | Plan for 3 or 4 emails |
| Hunter, State of Email Outreach 2026 (software vendor, no date shown, checked 20 October 2026) | 31 million emails sent by Hunter users in 2025 | 3 emails in total |
| Jed Mahrle, Practical Prospecting newsletter (agency founder, 14 June 2026) | The agency's client campaigns; no sample size published | 2 emails, then re-engage 60 to 90 days later |
All five describe their own users or clients, so none is neutral. Read them for the shape of the curve: where replies come from, and where an extra email stops adding much.
What share of replies comes from the first email?
Somewhere between two in five and three in five, depending on whose data you read. Instantly says 58% of replies in its 2025 data came from step one. A chart in the same report splits the rest: 23% from step 2, 12% from step 3 and 7% from step 4 onwards. Its headline figures also say 93% of replies arrive by step 3.
Belkins found nearly the reverse split. Across 7,530,489 emails from 2025, the first email produced 41.4% of replies and steps 2 to 6 produced 58.6%, even though the first email had the highest reply rate of any single step.
Both figures are shares of all replies, taken from different senders. A share like this depends on how many senders in the dataset run long sequences and how good their later steps are, so it moves from one platform to the next. Instantly doesn't say whether auto-replies are in its count. Belkins' companion reply-rate study, on the same 7,530,489 emails, excludes auto-replies and bounce notifications; the follow-up page doesn't restate the rule.
Reply rates for each step give a cleaner view, because they show what each email earns on its own:
| Email in the sequence | Smartlead: reply rate for emails sent at that step | Belkins: reply rate by step (2025) |
|---|---|---|
| 1 | 1.18% | 0.59% |
| 2 | 0.73% | 0.51% |
| 3 | 0.56% | 0.30% |
| 4 | 0.49% | 0.34% |
| 5 | 0.37% | 0.30% |
| 6 | 0.35% | 0.28% |
| 7 | 0.33% | Not reported |
Read each column on its own: the two come from different senders and datasets, so the numbers can't be compared across columns. Smartlead's falls fastest after the first email, Belkins' after the second. In Belkins' data step 4 edges above step 3, which Belkins reads as a sign that a fourth email with a genuinely different angle can beat a generic third one.
Do more follow-ups get more replies in total?
Up to about three emails, in the data that measures whole sequences. Hunter's 2026 report, based on 31 million emails its users sent in 2025, gives the reply rate per recipient by sequence length:
| Emails in the sequence | Hunter: reply rate |
|---|---|
| 1 | 3.3% |
| 2 | 6.1% |
| 3 | 6.8% |
| 4 or more | 4.1% |
Hunter says the best sequences used three messages in total, and that its previous year's analysis peaked at four. It reads the change as recipients being less willing to sit through long sequences.
Smartlead does the sum another way. If every contact received every step at its average step rates, a 3-email sequence would earn about 25 replies per 1,000 contacts, 52% of them from follow-ups, and a 7-email sequence about 40. Smartlead flags that this assumes everyone gets every step, and that its data can't measure annoyance, unsubscribes or spam complaints.
Meetings shift the picture slightly. In Belkins' data, step 3 booked 35.6% of email-sourced meetings, the largest share of any single step, and steps 3 to 5 together booked 53.5%. Belkins' text says step 3 books more than steps 1 and 2 combined, but its own figures give those two 46.4% combined (26.8% and 19.6%), more than step 3's 35.6%. It also says the appointment rate settles at 0.2% from step 3 onwards, twice that of steps 1 and 2, and puts this down to a filtering effect: people still in the sequence by then haven't opted out or said no. Belkins doesn't define that rate, and it is hard to square with the 1,200-plus appointments its companion study reports from 7.5 million emails.
Across the three, the second and third emails still earn replies and meetings. The case for a fourth or later email is mixed.
Why Jed Mahrle stops at two emails
Because a third email almost never moves the needle in the agency's campaigns, says Jed Mahrle, founder of Practical Prospecting, which runs outbound campaigns for B2B clients. In the newsletter of 14 June 2026, Mahrle wrote: "Two emails is almost always enough." The same tip suggests a strong first email, a short follow-up (maybe with an image), then a return 60 to 90 days later with a fresh angle.
Other issues explain the worry. In an April 2026 issue Mahrle wrote that people only notice and remember a sender negatively when 4, 5 or 6+ emails pile into one thread or sequence, and that clean sequences of 1 to 3 emails with good copy are not burning your total addressable market. The older templates were also short: a December 2023 issue lays out four emails over ten days in two separate threads, and notes that leaving out 20+ steps is intentional.
Mahrle publishes no sample size, so treat this as one experienced practitioner's view. It fits the numbers better than it first looks: in Hunter's data the gain from a second to a third email is small (6.1% to 6.8%), and in Belkins' data the reply rate falls most after the second email (0.51% to 0.30%).
How many LinkedIn messages should you send?
Three in total, in Expandi's data. Expandi's State of LinkedIn Outreach H2 2026 report grouped 2,966 campaigns created in 2026, each with at least 50 contacts, by how many messages they contained:
| Messages in the campaign | Campaigns | People contacted | Reply rate |
|---|---|---|---|
| 1 | 619 | 203,520 | 6.3% |
| 2 | 412 | 134,879 | 6.8% |
| 3 | 784 | 277,395 | 9.8% |
| 4 | 429 | 122,567 | 7.9% |
| 5 or more | 722 | 208,336 | 5.0% |
Three things to know before you use this table:
- The note counts. In Expandi's connector campaigns the first message is the note sent with the connection request, so three messages can mean a note plus two messages after someone accepts.
- It shows association. Expandi doesn't spell out the reply-rate formula for this table, and it describes its findings as associations, not experiments. For the drop at five or more, it suggests prospect fatigue, or long campaigns aimed at less receptive audiences.
- The second message works hardest. Measured per message, the second message earned 8.8% replies from the people who received it, the highest of any position, so Expandi suggests putting your strongest pitch there.
For combining channels, see our guide to multichannel outreach, and our LinkedIn follow-up message templates for wording.
Our recommendation: a three-touch plan for email and LinkedIn
Send three emails over about eight days in one thread, and on LinkedIn a connection request plus up to two messages, then stop on any answer. Use this plan on the next list you contact. The timings are our advice, set against reply-speed data: Smartlead says 93.9% of replies to an email arrive within 72 hours, and Instantly suggests spacing touches 3 to 4 days apart.
Day 1: the first email
Put most of the work here: it earns the highest single-step reply rate in both email datasets above that report step rates, Smartlead's and Belkins'. Keep it short: Smartlead found first emails under 50 words earned the highest reply rate, 1.47%.
Day 4: follow-up one, same thread
Add one new thing: a customer example, a screenshot or a sharper statement of the problem. Instantly says step-2 emails that read like a reply beat formal follow-ups by about 30%. For wording, start from our follow-up email templates.
Day 8: follow-up two, then stop
Use a different angle and a smaller ask. If this third email earns nothing across your first few hundred contacts, drop to two emails, as Mahrle does.
On LinkedIn: a request and two messages
Send the connection request with a short note, then up to two messages after the person accepts: three messages in Expandi's count. Make the first message after acceptance your strongest, and don't open by introducing yourself again. When someone accepts, Siela's AI SDR reads their profile again and picks up from the note instead of re-introducing itself.
Stop on any answer
Let a booking, a yes, a no or a request to stop end the sequence on every channel. In Siela, AI follow-ups continue in the same email thread and stop when the person books, says yes, says no or asks to stop.
Come back in a quarter
For people who never replied, Mahrle returns 60 to 90 days later with a fresh angle. Belkins suggests a similar final email with a low-commitment ask and a long time horizon, such as reconnecting next quarter.
Frequently asked questions
Why do Instantly and Belkins disagree on how many replies the first email gets?
They studied different senders. Instantly's 58% comes from users of its software in 2025, while Belkins' 41.4% comes from 7,530,489 emails in its own agency campaigns that year. A share of replies depends on how long the sequences in each dataset run and how good the later steps are. Per-step reply rates, which Belkins and Smartlead publish, are a fairer guide.
Should a follow-up email go in the same thread as the first email?
Usually, yes. Instantly's 2026 benchmark report says step-2 emails that read like a reply, such as a quick note pointing to the email below, beat formal follow-ups by about 30%. Jed Mahrle's published templates also keep each follow-up in the same thread as the email it follows, and start a new thread only for a new angle.
How many messages should a LinkedIn sequence have after the connection request?
Two, if you count the connection note as the first. Expandi's H2 2026 report found campaigns with three messages in total replied at 9.8%, against 5.0% for five or more, across 2,966 campaigns created in 2026. In its connector campaigns the first message is the connection note. Expandi describes these results as associations, not experiments.
When should you contact someone again after a sequence ends?
Wait at least a month, and bring a new angle. Jed Mahrle suggests re-engaging 60 to 90 days after a two-email sequence, and in an August 2025 issue described pausing and retargeting people a month later with new messaging. Belkins suggests a last email with a long time horizon, such as asking whether to reconnect next quarter.
Do long email sequences hurt deliverability?
None of the studies here measured it directly. Smartlead says its step data can't measure annoyance, unsubscribes or spam complaints, and Hunter's report gives no complaint figures by sequence length. Jed Mahrle argues that recipients remember and block senders who pile four or more emails into one sequence. Until you have your own numbers, watch unsubscribes and complaints at each step.
Sources
- Instantly: Cold Email Benchmark Report 2026 (updated 12 Jan 2026)
- Belkins: Sales follow-up statistics in B2B (2026 study) (updated 26 Jun 2026)
- Belkins: What are B2B cold email response rates? (2026 study) (updated 26 Jun 2026)
- Smartlead: How many follow-up emails is too many? (checked 20 Oct 2026)
- Smartlead: Do cold email follow-ups work? (checked 20 Oct 2026)
- Smartlead: How long should you wait before following up on a cold email? (checked 20 Oct 2026)
- Smartlead: Cold email statistics and benchmarks for 2026 (checked 20 Oct 2026)
- Hunter: The State of Email Outreach 2026 (checked 20 Oct 2026)
- Expandi: The State of LinkedIn Outreach H2 2026 (checked 20 Oct 2026)
- Jed Mahrle, Practical Prospecting #148: 14 Cold Email Tips That Actually Work (published 14 Jun 2026)
- Jed Mahrle, Practical Prospecting #145: Why You Should Send More Emails (published 19 Apr 2026)
- Jed Mahrle, Practical Prospecting #86: The Only Two Types of Emails You Need (published 10 Dec 2023)
- Jed Mahrle, Practical Prospecting #124: Why Burning Your TAM is a Myth (published 10 Aug 2025)
- Jed Mahrle, Practical Prospecting: Update (newsletter has a new home) (published 19 Jun 2025)