Key finding
Across 10,023 active account-days of LinkedIn outreach, 75.8% ran at 25 or more connection invites, and the median sending day and the 90th-percentile sending day are the same number: 25 invites, the platform ceiling.
What we measured
LinkedIn limits how many connection invites an account can send. What nobody outside a platform can see is how real accounts behave against that limit: whether teams pace themselves below it, or press against it every day they send.
This study measures it directly. An account-day is one LinkedIn account on one calendar day with at least one invite sent. Across 10,023 of them, we report the distribution of daily invite volume.
Methodology
Dataset window. All connection requests sent between 8 January 2025 and 27 August 2026, pulled on 27 August 2026, grouped by sending account and UTC calendar day. Unlike rate studies, volume needs no maturity restriction, so this uses the full window.
Denominator. Days with zero invites are not counted; every share below is a share of days on which the account sent at all. This is deliberate: the question is what a sending day looks like, not how often accounts rest.
Why account-days. Counting days rather than accounts describes operating behavior without exposing population counts, and it weights the answer by activity, which is the honest weighting for a question about how sending is actually done.
The findings
| Metric | Value |
|---|---|
| Active account-days | 10,023 |
| Mean invites per active day | 22.41 |
| Median | 25 |
| 90th percentile | 25 |
| Days at 20 or more invites | 8,288 (82.69%) |
| Days at 25 or more invites | 7,594 (75.77%) |
The distribution is not a curve, it is a wall. When the median and the 90th percentile of a distribution are the same number, the distribution has collapsed onto that number. Three out of four sending days ran at 25 or more, and the mean of 22.41 is dragged below the median only by a minority of light days.
A data note rather than a finding: 93 days (0.93%) recorded more than 25 invites, with one recorded at 328. These are attribution artifacts in day boundaries and account bookkeeping, not a way around the platform's limits, and we report them rather than trimming them.
What this means
The cap is not a boundary teams approach carefully. It is the default operating point. Whatever daily allowance exists gets spent, nearly every day, by nearly everyone sending.
That collides with a finding from this platform's earlier frozen cohort. The Volume Tax measured acceptance by average daily volume and found the 20-29 invite band accepting 5.4 points below the 10-19 band. This study measures how many days are actually in that heavy band: three quarters. The worst-measured operating point and the near-universal operating point are the same place. If the volume tax is real, almost everyone is paying it, and almost nobody has tried the alternative at scale: the lighter bands are small, and nothing in this data can show they are deliberate pacing rather than ramp-ups and partial days.
There is a rational reading and an irrational one. The rational: an account is a fixed cost, and running it below capacity wastes the cost regardless of per-invite efficiency. The irrational: nobody is deciding at all, because tools default to the maximum and defaults win. The data cannot separate the two, but it can say the choice is nearly unanimous.
In practice
An SDR configuring a new account's daily limit is not really choosing a number; they are choosing whether to accept the default everyone else runs. In this data, picking 15 a day would put their sending days in a band occupied by 17.3% of account-days, and the data cannot distinguish how much of that band is deliberate pacing rather than ramp-ups and partial days.
A RevOps lead modeling capacity can treat 25 invites per active account-day as the planning number: the median day is 25, the mean across all active days is 22.41, and the gap between the two is the rate of partial days, not variation in the ceiling.
Limitations
One platform's users. Everyone in this dataset chose an automation platform, which selects for volume-oriented teams. Hand-senders and dabblers are underrepresented, and a population of casual users would show a different distribution.
Account-days, not accounts. Heavy accounts contribute more days, so this weights by activity. That is the intended question, but it means the 75.77% describes days, not senders.
The ceiling itself is contextual. The 25-a-day operating ceiling reflects the platform limits and safety pacing in force during the window. If LinkedIn changes enforcement, the wall moves, and this snapshot dates itself.
Artifact days. The 0.93% of days above 25 are disclosed above; removing them entirely would not change any headline figure by a visible amount.
Concentration. As across this dataset, sending is concentrated: the largest workspace accounts for 22.88% of matured requests. The wall shape holds with or without any single workspace, but per-band shares move by points, not the picture.