Quick answer: Across 5,685 people who were sent a first LinkedIn message from 22 accounts between 2 June and 13 October 2026, 12.7% replied within a week. The middle account landed at 12.1%, and the middle half of accounts between 7.2% and 17.2%. But only about 1 in 6 replies showed real interest, and a polite "thanks" was the single most common answer. Replies come fast (66% within a day, 90% within a week), and follow-ups still earn answers: 19% of the people who replied did so after a second, third or fourth message.
Most reply-rate advice is a single number with no definition. These are measured, with the definition, and with the places where the data cannot tell you anything.
How these numbers were counted
- What counted as a first message: the earliest LinkedIn direct message a FollowArc account actually sent to a person (not one that was queued, skipped or blocked). Each person was counted once per account.
- When it was measured: only people whose first message was at least 7 days old, so nobody was judged before they had a fair chance to answer. First messages were sent between 2 June and 13 October 2026.
- What counted as a reply: the person wrote back after that first message. Any answer counts, including "no thanks" and "thanks". A reply is counted as "within a week" only if it arrived inside seven days of the first message.
- Where it comes from: 22 FollowArc accounts (a person, team or client each) across 73 campaigns. Test and demo workspaces were left out. No account, person or company is identified here.
| Measure | Result |
|---|---|
| People sent a first message | 5,685 |
| Replied within 7 days | 723 |
| Overall reply rate | 12.7% |
| Accounts with at least 100 people | 19 |
| Lower quartile (25th percentile) | 7.2% |
| Median account | 12.1% |
| Upper quartile (75th percentile) | 17.2% |
| Highest account | 22.7% |
One account recorded no replies at all. We cannot tell from the data whether that is a true zero or a gap in reply tracking for that account, so we would not read anything into the low end. We would read the middle.
What is a good LinkedIn reply rate?
In this data, about 12% is the middle for a first message to someone who has just accepted your request. Treat roughly 7% as the floor for ordinary outreach and 17% or more as what a well-matched audience and a short message can reach. This is a reply rate to the first message, not a meeting rate: the section below on what the replies said is the part that matters for pipeline. For a longer explanation of what counts as a good rate, see what is a good LinkedIn reply rate.
What did the replies actually say?
This is the finding most outreach reporting hides. Of the 723 people who replied within a week, our reply classifier sorted them like this:
| What the reply was | Share of replies |
|---|---|
| A polite acknowledgment ("thanks", "good to connect") | 41.1% |
| Not interested | 18.1% |
| Interested | 17.6% |
| Other | 12.4% |
| Not classified | 4.6% |
| Maybe | 3.3% |
| An objection | 2.2% |
| Out of office | 0.6% |
| Asked to be removed | 0.1% |
The classification is automatic, not hand-checked, so treat the shares as approximate. The point stands: a 12.7% reply rate becomes roughly 2.2% of people showing clear interest (127 of 5,685), about 2 in every 100 people messaged. If you plan pipeline from the headline reply rate, you will overestimate it by almost six times. To turn your own rates into expected meetings, use the free outreach ROI calculator.
How fast do people reply?
Quickly. Counting every person who replied at any point after the first message (800 people):
| Replied within | Share of replies |
|---|---|
| The first day | 66% |
| 3 days | 82% |
| 7 days | 90% |
The median reply arrived about 5 hours after the message. That is why a seven-day window is a fair way to judge a first message, and why a message sent yesterday has hardly been tested.
Do follow-ups work?
Yes, a little less each time, and they are worth sending. The table shows, for each message in a sequence, how many people who had not yet replied wrote back within three days of that message and before the next one went out.
| Message in the sequence | Sent | Replied within 3 days | Reply rate |
|---|---|---|---|
| First message | 5,656 | 538 | 9.5% |
| Second | 1,156 | 93 | 8.0% |
| Third | 430 | 31 | 7.2% |
| Fourth or later | 174 | 7 | 4.0% |
Looked at from the other side, of the 800 people who replied at all, 81% replied after the first message, 13% after the second, 4% after the third and 1% after a fourth or later. So most of the value is in the first message, but about one reply in five only came because someone followed up. A second message gets close to the first one's response, which is a stronger case for a short follow-up than for stopping after one. Wording ideas are in LinkedIn follow-up message templates, and how to handle people who go quiet is in following up with prospects who went quiet.
Does it matter how soon you message after someone accepts?
It did. This comparison only includes people who accepted a connection request we sent, so a warm list of people who were already connected cannot inflate it (in our first pass it did, which is why this check matters). 3,414 people qualified.
| First message sent after acceptance | People | Replied within 7 days |
|---|---|---|
| Within 1 hour | 940 | 21.1% |
| 1 to 24 hours | 1,737 | 17.0% |
| 1 to 3 days | 301 | 14.6% |
| 3 to 7 days | 186 | 9.1% |
| More than 7 days | 250 | 9.6% |
The raw gap is large, but different accounts message at different speeds, so we also compared the same account against itself. In the 10 accounts with at least 40 people in both groups, messaging within 24 hours beat messaging later in 8 of 10, by an average of 5.1 points (median 4.4). That is a smaller and more believable effect than the raw table. It is not a controlled test: the people who wait longest may also be the ones whose message was held by a cap or a schedule for other reasons.
In this data the typical wait was 9 hours, 78% of people were messaged inside a day, and 13% waited three days or more. If your first message goes out days after acceptance, that tail is the part to fix.
Does message length matter?
Long messages did worse. Raw reply rates by length of the stored message text:
| First message length | People | Replied within 7 days |
|---|---|---|
| Under 150 characters | 1,380 | 14.3% |
| 150 to 299 | 470 | 12.6% |
| 300 to 499 | 2,262 | 15.9% |
| 500 or more | 1,573 | 6.8% |
Within the same account, a message under 500 characters beat one of 500 or more in 9 of 10 accounts (those with at least 40 people in each group), by an average of 4.7 points. Below 500 there was no clear winner, so we would not read the 12.6% for the 150 to 299 group as a rule against mid-length messages. The length is of the message as written in the campaign, before personalisation was filled in, so real sent lengths run a little different. The practical reading is simple: keep the first message short and make it one clear ask. Examples are in how to write cold LinkedIn messages that get replies and our free cold message clarity checker.
Does a link in the first message hurt?
We cannot say. Across all accounts, messages with a link had a much lower reply rate in the raw numbers, but only one account had enough messages both with and without a link to compare fairly, and one account is not evidence. We would rather say that than publish a raw gap that mostly reflects which accounts use links. If you want to know for your own audience, test a version with and without a link, in the same account, with the same audience.
Does the weekday matter?
Hardly. Replies within three days of a first message sent Monday to Friday ranged from 11.0% to 12.4%. Weekends were lower (10.2% Saturday, 9.4% Sunday) on far fewer messages, so treat that as a hint rather than a finding.
| Day sent (Central Time) | People | Replied within 3 days |
|---|---|---|
| Monday | 1,145 | 12.4% |
| Tuesday | 929 | 11.4% |
| Wednesday | 902 | 11.0% |
| Thursday | 972 | 12.1% |
| Friday | 905 | 11.5% |
| Saturday | 608 | 10.2% |
| Sunday | 224 | 9.4% |
What these numbers do not tell you
- They are not a random sample of LinkedIn. These are accounts run through one platform, many of them agencies and consultants with defined target lists. A cold, broad audience will usually do worse.
- A reply is not a meeting. Roughly one reply in six showed clear interest, and we did not follow every conversation to a booked call here.
- The classification is automatic. The split between acknowledgments, interest and refusals is approximate.
- Only the within-account comparisons are fair tests, and even they are not experiments. Speed and length were not randomly assigned.
- We did not control for audience or offer. They are the largest unmeasured factors.
- Small groups are noisy. The weekend figures and the 4+ follow-up row rest on few messages.
How to use this
- Measure your own baseline over 7 days before comparing yourself to any number here. Compare yourself to yourself first.
- Judge a first message by clear interest, not by replies. Plan pipeline from the 2-in-100 figure, not the 12.7%.
- Message within a day of acceptance. Do not let the first message wait for a weekly batch.
- Keep the first message under 500 characters and ask one thing.
- Send at least one follow-up, and judge each message only after about three days.
- Test one change at a time inside the same account. It is the only comparison that holds up.
For how many connection requests a day is safe, see how many LinkedIn connection requests per day. For the stage before this one, see our acceptance-rate benchmarks from 25,879 invites.
To see where your own acceptance and reply rates sit among these accounts, use the free outreach benchmark checker.
Where FollowArc fits
FollowArc runs the outreach these numbers come from: connection requests, follow-ups and replies, with sending paced inside configured windows. Its agent, Sophia, ranks replies by intent so a real expression of interest is easy to tell from a polite acknowledgment. You can try FollowArc free for 7 days and measure your own baseline.
Figures are from FollowArc data, 2 June to 13 October 2026, and are aggregated and anonymous. We will refresh them as more messages mature.