benchmark · · 10 min read
LinkedIn cold outreach benchmark 2026: acceptance, reply and follow-up rates from one account, updated weekly
One account, 158 connection requests to strangers since September 15, 2026: 55% of the requests old enough to judge were accepted, against a platform average of 28.5%. Time to accept, time to reply, follow-up yield, decisions per day and the week-by-week table, refreshed every week with the method behind every number.
by Thiago Valentim · Scout With Me, by Worth Contacting · updated
This page holds every rate measured on this blog in one place, so there is one address to check and one to cite. The numbers come from a single LinkedIn account, mine, prospecting for real: strangers chosen after reading their profile, one connection request each with a note written from that profile, and the replies read by hand. Each section says which batch it comes from and when it was counted, and the changelog at the end says what changed. It is updated every week.
What is a good LinkedIn connection acceptance rate in 2026?
Above a half, on this account, once a batch has had time to mature. Of the 102 requests sent at least two weeks ago, 56 were accepted. The largest public dataset puts the platform-wide average at 28.5%.
| This account | Platform-wide | |
|---|---|---|
| Connection requests | 158 | 13,218,869 |
| Accounts | 1 | 13,302 |
| Period | Sep 15 to Oct 14, 2026 | May 2025 to Apr 2026 |
| Accepted, requests at least two weeks old | 55% (56 of 102) | |
| Accepted, all requests, including the 25 sent in the last two days | 47% (75 of 158) | 28.5% |
| Replied to the request note | not sent as a question | 3.0% |
| Replied to a message after accepting | 63% of accepted (people who wrote at least once) | 10.4% (of messages) |
Counted on October 14. The platform figures are from Expandi's 2026 outreach benchmark, which mixes every kind of request its users send. Every request on this account carried a note written from the person's profile and was approved by me before it went out, which is the most likely reason for the gap. The two reply figures measure different things, people here and messages there, so read them as a direction, not a ratio.
How many replies do 100 connection requests produce?
On this account, 100 requests to strangers become 54 accepted, 34 people who write back, and 19 who keep the conversation going. From the 133 requests sent between September 15 and October 9, all at least five days old:
| Step | People | Share of requests |
|---|---|---|
| Connection requests to strangers | 133 | 100% |
| Accepted | 72 | 54% |
| Replied at least once | 45 | 34% |
| Wanted to talk (wrote back twice or more, did not say no), as read on October 12 | 25 | 19% |
| Declined politely or asked not to be contacted | 5 | 4% |
Counted on October 14. The full funnel, from 193 profiles read down to the 25, is in Can you get customers on LinkedIn without an audience?
How long does a connection request take to be accepted?
Half within about 14 hours, almost all within a week. Of the 74 acceptances with a recorded time, counted on October 14:
| Accepted | People | Share | Cumulative |
|---|---|---|---|
| Within 24 hours | 50 | 68% | 68% |
| Day 2 to day 3 | 8 | 11% | 78% |
| Day 4 to day 7 | 12 | 16% | 95% |
| After a week | 4 | 5% | 100% |
Median 0.6 days, with a quarter of acceptances inside five hours and three quarters inside 1.2 days. The rule for pending requests: How long does a LinkedIn connection request take to be accepted?
How long until the first reply?
Half within 7 hours. Of the 47 people who replied, counted from my first message after they accepted:
| First reply | People | Share |
|---|---|---|
| Within 24 hours | 37 | 79% |
| Later | 10 | 21% |
Median 6.7 hours; a quarter within 40 minutes, three quarters within 22 hours.
When do people reply?
In working hours, on working days. The 47 first replies by time of day and day of the week, local time of the account (Eastern European Summer Time):
| Time of day | Replies | Day of the week | Replies | |
|---|---|---|---|---|
| Morning, 6 to 12 | 20 | Monday | 7 | |
| Afternoon, 12 to 18 | 15 | Tuesday | 11 | |
| Evening, 18 to 24 | 11 | Wednesday | 9 | |
| Night, 0 to 6 | 1 | Thursday | 8 | |
| Friday | 10 | |||
| Saturday | 1 | |||
| Sunday | 1 |
Two of 47 replies came on a weekend. A message sent on Friday afternoon is read on Monday.
Do follow-ups work on LinkedIn?
The first one does. The second one produced one reply in seventeen. Of everyone who got each message, counted on October 14:
| Message | People who got it | Replied to it | Rate |
|---|---|---|---|
| First message after accepting | 75 | 34 | 45% |
| First follow-up | 37 | 12 | 32% |
| Second follow-up | 17 | 1 | 6% |
Counting only the people who had a full week to answer: 10 of 28 replied to the first follow-up, 0 of 7 to the second. One more person wrote before my first message went out. What each message added, and when to send the follow-up: Do LinkedIn follow-ups work? How the replies read, tone by tone, is in Is it rude to cold message strangers on LinkedIn?
Does recent posting predict who accepts and replies?
Yes, by about two to one on acceptance and three to one on replies. From the first 102 requests, split by what the profile's activity tab showed when it was read, counted on October 8:
| Last visible post | People | Accepted | Replied |
|---|---|---|---|
| Within the last 30 days | 80 | 53% | 29% |
| Older than 30 days | 18 | 28% | 11% |
Four profiles with no visible activity section are left out. The rule this produced: Does posting on LinkedIn predict who replies?
What does a week of requests look like?
Calendar weeks, Monday to Sunday. Acceptances count in the week the request was sent; replies count in the week the first message was sent. Counted on October 14:
| Week of | Requests | Accepted | First messages | Replied |
|---|---|---|---|---|
| September 14 | 53 | 25 (47%) | 17 | 12 (71%) |
| September 21 | 17 | 8 (47%) | 12 | 10 (83%) |
| September 28 | 37 | 27 (73%) | 27 | 15 (56%) |
| October 5 | 26 | 12 (46%) | 16 | 8 (50%) |
| October 12, three days in | 25 | 2 (8%) | 3 | 2 |
The last row has not had its week yet; its acceptances and replies arrive over the next seven days. A new row is added each week.
What do strangers do when they reply?
Of the 43 who had replied by October 12, read by hand, one category per person:
| What the person did | People |
|---|---|
| Kept the conversation going | 31 |
| Asked what I wanted, without declining | 5 |
| Said the first message sounded automated | 3 |
| Declined politely | 3 |
| Asked not to be contacted | 1 |
| Rude, angry or offensive | 0 |
Five more people replied for the first time between October 12 and 14: three answered and carried on, one was wary of the motive, one sent a stock greeting and little else. None was rude. Sixteen of the 43 exchanged three or more messages; two wrote ten.
How much time does it take per day?
From September 16 to October 14, 287 decisions were mine: 204 drafts approved, 68 sent back for another version, 2 replies discarded, 8 fit calls corrected, 5 conversations taken over. There was something to decide on 24 of the 29 days, a median of 9 decisions on those days and a maximum of 40. Decisions come in short sessions: the typical one is 3 or 4 decisions in under two minutes, about half a minute each; the longest was 21 decisions in 18 minutes. The arithmetic and the sessions, one by one: How much time does LinkedIn prospecting take per week?
How to use this benchmark
- Plan per hundred: on a list of a hundred strangers who fit, expect about fifty acceptances, thirty replies and twenty conversations. Half that if the notes are generic.
- Judge a batch after a week, not a day. Day two shows about two thirds of the acceptances you will have on day seven.
- Send one follow-up, a few days later. Not a second.
- Sort the list by the date of the last post before spending the week's requests.
- Send in working hours; two of 47 replies came on a weekend.
- Expect silence from most and anger from none.
Scout does the finding, the reading and the drafting on your computer, on your account, and keeps these counts for you; deciding, replying and stopping stay yours.
How these numbers were measured
- Source: Scout running on my own LinkedIn account, prospecting for real, for the products I sell. No data from other users of Scout; the app keeps its data on each user's computer and sends nothing to us. The platform-wide figures are Expandi's, linked above, and are theirs alone.
- What was counted: connection request, sent with a note written from the profile, to someone who was not a connection; accepted, the request was accepted; replied, the person wrote at least once after accepting; wanted to talk, wrote at least twice and did not say no; follow-up, how many of the people who got that message replied after it and before the next one; tone, one category per person, read by hand across all of their messages; decision, an action of mine recorded by the app with its time.
- Batches: the comparison, the timings, the follow-ups, the weekly table and the decisions use all requests from September 15 to October 14, counted on October 14. The hundred-request funnel uses the 133 requests of September 15 to October 9. The posting split uses the first 102 requests, counted on October 8. The tone table is as read on October 12.
- Windows: a batch is judged after one week for acceptances and two for replies; the last week's requests are inside both windows, so their numbers can still grow.
- Size and rounding: rates are rounded to whole percentages and shown with their base. Groups smaller than ten people get no row.
- Left out: people who were already connections; the 60 profiles read and not messaged in the first batch; nobody is named, no message is quoted.
- The rules that apply to every post are in How the numbers on this blog are measured.
This is a living page: every week the new requests are added, the counts move, and the changelog records it.
Changelog
- 2026-10-13: first version, assembled from the three experiments published October 7 to 12 and the platform-wide comparison.
- 2026-10-14: weekly recount. Requests 133 to 158, acceptances 72 to 75, replies 43 to 47. Acceptance rate now shown for matured requests (55% of 102) beside the all-requests figure (47%). Time to accept refreshed on 74 acceptances, time to reply on 47. New sections: when people reply, by hour and weekday; decisions per day. Follow-ups now count everyone who got each message and add the one-week maturity view. The weekly table switched to calendar weeks with replies counted by first-message week.