The most persistent debate in B2B prospecting is how many follow-ups are too many. Do you stop after two unanswered messages, or push on to a third and fourth?
Here's the short answer the public data supports: the first message usually collects the largest share of replies, because everyone receives it. But follow-ups add up to a big chunk of the total, and the message with the highest reply rate per send isn't always the first one. How much each touchpoint is worth varies a lot from source to source, and very little of it has been measured for LinkedIn prospecting as carefully as it has for cold email.
This guide pulls together what the public sources actually show, explains why they seem to contradict each other, flags what nobody has measured (voice notes, profile views, liking a post before you message), and walks through how to find out which touchpoint works in your own cadence.
Which cadence touchpoint gets the most replies?
Connection request notes and second follow-up messages get the highest reply rates on LinkedIn. According to Expandi's analysis of 6.7 million messages, the second message in a sequence achieves the highest reply rate per send (8.8%), outperforming both single-message blasts and longer sequences.
For the share of total replies each touchpoint accounts for, the public numbers come from cold email, not LinkedIn:
| Source | What was measured | What it says about touchpoints |
|---|---|---|
| Expandi (2026) | 13.2M connection requests and 6.7M LinkedIn messages across 13,302 accounts on its platform, May 2025 to April 2026 | Message 2 has the highest reply rate per send: 8.8%. In connector campaigns, "message 1" is the connection-request note |
| Belkins (2026) | 7,530,489 cold emails sent by the agency's managed outreach over 12 months in 2025 | Email 1 has the highest per-step rate (0.59%) but captures only 41.4% of replies; steps 2 through 6 account for 58.6% |
| Instantly (2026) | Cold email on its platform, January 1 to December 18, 2025 | Step 1 generates 58% of replies; follow-ups, 42% |
| Backlinko and Pitchbox (2019) | 12M content and SEO outreach emails (link building, guest posts, reviews, sponsorships) | A single follow-up boosts replies by 65.8% versus no follow-up |
Three caveats change how you should read that table.
Two different questions are getting mixed up. "Which touchpoint collects the most replies?" and "which touchpoint has the highest reply rate?" are not the same question. The first message goes to everyone; the third follow-up only goes to people who haven't replied yet. So the first message tends to collect more replies in absolute terms, even when a later touchpoint converts better per send. Expandi puts it this way: the opening message "collects the most raw replies because everyone receives it," but messages 2 and 3 convert at a higher rate per person reached.
The denominator changes from source to source. Expandi divides replies by messages sent. Belkins measures replies per email sent at each step. Instantly and Belkins also report each step's share of total replies, which is a different calculation. An 8.8% rate from one source and a 0.59% rate from another aren't comparable: one is LinkedIn after a connection is accepted, the other is cold email to someone who has never heard of you.
The sources disagree on size. At Belkins, the first email gets 41.4% of replies; at Instantly, 58%. Both measure cold email in the same year, and the first touch's share still differs by more than 16 points. Any single figure along the lines of "X% of replies come from touchpoint 1" describes one dataset, not a law of the channel.
For a general sense of acceptance and reply rates, see the global LinkedIn prospecting benchmarks, read with the same caution about what each number measures.
The first message after acceptance
In Expandi's data, the message with the highest reply rate per send is the second one in the campaign. That sounds counterintuitive until you look at how the campaigns are built: in connector campaigns, "message 1" is the note attached to the connection request, and recipients can't reply until they accept. "Message 2" is, in practice, the first message after the connection is accepted.
In other words, the finding backs up something prospecting teams already do: the first message after someone accepts your request is the most important touchpoint in the cadence. Expandi recommends putting your strongest pitch there, not in the connection note.
The same report has another useful number for thinking about the note: the average reply rate to the connection-request note is 3.0% (replies to the note divided by connection requests sent), versus 10.4% for outbound messages. The note decides who gets into the conversation; the conversation starts afterward. For when to add a note and how to write one, see the guide on connection acceptance rates and notes.
Why the first post-acceptance message performs best is a hypothesis, not a measurement: accepting is a sign of recent attention, and the message lands while the person still remembers who sent the request. Expandi describes it as a context where replying is "frictionless." No public source has measured how long that window lasts.
What about follow-ups? How many messages should you send?
Here the email sources and the LinkedIn source agree on the shape and disagree on the number.
On LinkedIn, Expandi compared 2,966 campaigns with at least 50 contacts each, grouped by number of messages:
| Messages in campaign | Campaigns | 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+ | 722 | 208,336 | 5.0% |
Source: Expandi, State of LinkedIn Outreach H2 2026. Campaigns created in 2026.
For cold email, Belkins calls 3 to 5 steps "the practical sweet spot for most B2B email sequences," and Instantly says 4 to 7 touchpoints. Belkins also reports meetings: steps 1 and 2 together account for 46.4% of email-sourced meetings booked, and steps 3 to 5 for 53.5%.
Before you turn that into a rule, three caveats:
It's association, not causation. Expandi states that all of the report's findings "are observational: they show association, not proof of causation." The company itself offers two explanations for the drop in campaigns with five or more messages: prospect fatigue, or self-selection, since campaigns that need five touchpoints may be targeting less receptive audiences. A study like this can't tell you how many replies your third follow-up will add.
Nobody knows whether the reply would have come anyway. Someone who replies at the third touch might have replied without it, a few days later. Only a test with a holdout group separates the two, and none of the sources above ran one.
Email isn't LinkedIn. The "share of replies by step" figures are cold email. LinkedIn's inbox is shorter and more personal, and a fifth unanswered message weighs more there. The drop for five-plus-message campaigns in Expandi's data is consistent with that.
The practical takeaway the sources support: stopping after a single message leaves replies on the table, and stretching the cadence indefinitely doesn't help. On LinkedIn, roughly three messages after acceptance is a reasonable starting point to test from. For more on structuring the sequence, see the LinkedIn B2B prospecting cadence guide.
What nobody has measured
Several touchpoints that show up in LinkedIn cadences have no reliable public data:
- Voice notes. We didn't find a study with a published sample and method comparing voice notes with text in prospecting.
- Profile views, likes, or comments before the request. Expandi says the top-performing connection notes reference an existing tie and recommends warming up before you connect, but it doesn't publish the isolated effect of each action.
- How much of the first-message reply is driven by the note. The previous version of this page gave a percentage for this. It had no source and has been removed.
- Breakdowns by role, industry, or deal size. The sources publish platform averages. Expandi shows that reply rate varies little by sender seniority and a lot by industry, but there's no "which touchpoint" cut by segment.
If someone quotes a number for any of these without saying where it came from, be skeptical.
Spacing between touchpoints
There's no public study measuring the ideal gap between LinkedIn messages either. What does exist is LinkedIn's own setting for recruiter InMail: an automatic follow-up can be scheduled 3 to 28 days after the first message, defaults to 7 days, and only goes out if the candidate hasn't replied. That's a product default, not a reply-rate measurement, but it's a sensible starting point.
In practice, a common and defensible setup:
- First message: a day or two after acceptance, so it doesn't read as an auto-reply.
- Follow-up 1: about a week later, with a complementary angle.
- Follow-up 2: one to two weeks later, with something new and useful that doesn't depend on the person remembering earlier messages.
- Close: if you use a fourth touch, make it clear it's the last one and make it easy to reply either way.
Those intervals are practice, not data. Test them.
What to say in later touchpoints
The sources repeat the same advice: every follow-up needs to bring something new. Instantly's first best practice for follow-ups is that each one adds new value or a new angle, "never just 'checking in.'" Belkins notes a slight uptick at step 4 over step 3 (0.34% versus 0.30%) and reads it as a sign that a well-crafted fourth touchpoint with a genuinely different angle can outperform a generic third. That's the agency's reading of its own data, not a controlled test.
The value follow-up. Give the person something they can use even if they never buy from you: a public figure from their industry with the source, an observation about something their company published, a specific question about a likely problem.
"Saw you opened SDR roles this month. When teams double in size, the first thing that tends to break is the lead handoff between SDRs and AEs. Is that on your radar?"
The gentle close. Signal that it's your last message, with no pressure.
"Last message from me here, so I don't clutter your inbox. If it ever makes sense to talk, just reply. If the timing isn't right, no problem at all."
No public data measures the reply rate of this format. Use it because it's respectful and ends the cadence cleanly, and measure the result in your own pipeline. For the case for following up at all, see why B2B follow-up on LinkedIn matters.
How to measure it in your own outreach
If you want to know which touchpoint gets the most replies from your audience, only your own data will tell you. There are two measurements, and they answer different questions.
1. Descriptive: where your replies come from. For every conversation with a reply, record which touchpoint it came in on. Calculate two things per touchpoint:
- share of replies: replies at that touchpoint ÷ total replies in the campaign;
- rate per send: replies at that touchpoint ÷ messages sent at that touchpoint.
The first shows where replies cluster; the second shows how much each touchpoint yields among the people who received it. Use a fixed attribution rule (for example, a reply counts toward the last touchpoint sent before it) and keep it the same throughout.
2. Holdout test: does the extra touchpoint add replies? The descriptive view can't tell you whether people who replied at the third touch would have replied anyway. For that:
- randomly split the leads who reached the decision point (for example, those who didn't reply to the first two touches) into two groups;
- one group gets the extra touchpoint, the other doesn't;
- compare the cumulative reply rate of both groups after the same window, say 14 days after the last touch of the group that got more messages.
Change only the number of touchpoints: same list, same account, same copy in the shared messages, same time period.
3. Respect sample size. Using the standard two-proportion calculation (95% confidence, 80% statistical power), in a hypothetical example:
- if the cumulative rate without the extra touch were 8% and rose to 12% with it, you'd need about 880 leads per group;
- if it rose from 8% to 10%, you'd need about 3,200 leads per group.
Those figures are arithmetic, not market data: swap in your own baseline and rerun the numbers in a sample size calculator. The lesson is that the lift from a late touchpoint, usually small in percentage points, only shows up at real volume. With a few hundred leads, a one- or two-point difference is noise.
4. Look at reply quality too. "Please remove me" is a reply. Split positive, neutral, and negative replies, and check whether the negative share grows at later touchpoints. A touch that adds replies and also adds "stop messaging me" may not be worth it.
5. Decide based on what shows up, including "nothing." If the extra touch doesn't visibly change the cumulative rate, cut it and put the time into the first message after acceptance, which is where most replies tend to land.
Where Chattie fits
Chattie doesn't publish platform usage data, so this guide doesn't quote any "Chattie" rates. What we can describe is how the product handles a cadence, because that's configuration, not statistics:
- Three sending modes: in Autopilot, the AI sends the message; in Semi-pilot, the AI sends until the lead replies and then a human approves each response; in Copilot, the AI writes it and a person approves it before it goes out. You can switch modes mid-conversation.
- Automated in all three modes: sending connection requests, campaigns and withdrawing pending requests after 15 days. The scheduling link goes out in the conversation: the AI sends it in Autopilot; in Semi-pilot and Copilot, with your approval.
- Fixed daily caps: 25 connection requests, 40 conversations, and 200 LinkedIn actions per day, with sending only during business hours. Those caps are fixed by default — above 25 connection requests per day, only with the user's own risk consent — and they're what keeps account-restriction risk low (never zero).
The caps also help you plan the test: at up to 25 connection requests a day, building a few hundred leads per group who reached the third touchpoint takes weeks on a single account. That's one more reason to test one question at a time and decide with your own numbers.
FAQ
Which message in a LinkedIn cadence is most likely to get a reply?
It depends on how you count. The first message tends to collect the most replies in absolute terms, because everyone receives it. By reply rate per send, Expandi's report on 6.7 million LinkedIn messages points to the second message in the campaign (8.8%), which in connector campaigns is the first message after acceptance.
Is it worth sending a 3rd and 4th follow-up on LinkedIn?
The third, probably yes; the fifth and beyond, probably not. In Expandi's data, three-message campaigns had a 9.8% reply rate, versus 6.8% with two, 7.9% with four, and 5.0% with five or more. That data is observational: it doesn't prove the extra touch causes the reply. Run a holdout test before locking in a rule.
How many follow-ups can I send on LinkedIn without being annoying?
There's no proven number. Expandi found the best results with three-message campaigns and a drop from five onward. For cold email, Belkins recommends 3 to 5 steps and Instantly 4 to 7. Spacing touches about a week apart and bringing something new each time matter as much as the count.
What share of replies comes from follow-ups?
In cold email, Belkins measured 58.6% of replies at steps 2 through 6, and Instantly 42% from follow-ups. The two sources differ by more than 16 points. There's no equivalent public figure with an open method for LinkedIn; measure it in your own outreach.
Do voice notes or profile views increase replies?
We didn't find a public study with a published sample and method that measures the isolated effect of voice notes, profile views, or liking a post before you message. If you want to use those touches, treat them as a hypothesis and test with a holdout group.
How do I know if my cadence is performing well by touchpoint?
Measure the reply rate per touchpoint (replies ÷ sends at that touchpoint) and each touchpoint's share of total replies, splitting positive and negative replies. Compare against your own history, not against other platforms' averages, which use different denominators.
Conclusion
"Which LinkedIn touchpoint gets the most replies?" has an honest two-part answer. The first message after acceptance is the most important touch: it collects most of the replies and, in Expandi's data, has the highest rate per send. Follow-ups still add replies, and roughly three messages perform best in the largest public LinkedIn dataset, but campaigns with five or more messages do worse.
What to do with that: write your first post-acceptance message with your strongest pitch, make follow-ups that bring something new instead of "just checking in," space touches about a week apart, and don't lengthen the cadence without testing. If you want to know what the third touch is worth for your audience, randomize a holdout group and decide with your own number.
If you want to run your cadence with daily caps that don't go up, and choose between the AI sending or you approving each message, take a look at Chattie, AI SDR for LinkedIn. To fold send day into your test, see also what we know about LinkedIn reply rates by day of the week.
References
Outside sources cited in this article, in the order they appear.
- Expandi — "The State of LinkedIn Outreach: Benchmarks Every GTM Leader Needs on Sequencing, Send Times, and AI vs. Human Messaging" (H2 2026): reply rate by message position, messages per campaign, note that findings are observational (https://expandi.io/state-of-linkedin-outreach-h2-2026/)
- Expandi — Stefana Zarić, "LinkedIn Outreach Benchmarks 2026," May 19, 2026: metric definitions, 28.5% acceptance, 3.0% note reply, 10.4% message reply (https://expandi.io/blog/linkedin-outreach-benchmarks-2026/)
- Belkins — Michael Maximoff, "Sales follow-up statistics in B2B: Belkins' 2026 study," updated June 26, 2026: per-step reply rate, share of replies and meetings by step across 7.5M emails (https://belkins.io/blog/sales-follow-up-statistics)
- Instantly — "Cold Email Benchmark Report 2026," updated January 12, 2026: 58% of replies on step 1, 4 to 7 touchpoints (https://instantly.ai/cold-email-benchmark-report-2026)
- Backlinko — Brian Dean, with data from Pitchbox, "We Analyzed 12 Million Outreach Emails. Here's What We Learned," April 16, 2019: effect of one follow-up (https://backlinko.com/email-outreach-study)
- LinkedIn Help — "Schedule a message or follow-up in Recruiter and Recruiter Lite": automatic InMail follow-up between 3 and 28 days, 7-day default (https://www.linkedin.com/help/recruiter/answer/a1457303)
See also
- Best AI SDR platforms for B2B SaaS
- B2B follow-up on LinkedIn: complete guide
- B2B prospecting cadence flow: what it is and how to build one
- LinkedIn prospecting benchmarks (global)
- LinkedIn prospecting benchmarks for Brazil
- How to increase your LinkedIn reply rate
- LinkedIn Connection Acceptance Rate: When to Add a Note and How to Write It