How many LinkedIn invites per day is actually safe?

What LinkedIn actually publishes, checked rather than repeated

No number. We went and read the help pages on 24 September 2026 rather than repeating the category, and LinkedIn states no invitation limit anywhere: not per day, and not the roughly 100 per week that nearly every vendor page asserts as fact. Its invitation-restriction page describes triggers in words only, naming invitations "sent many invitations within a short amount of time" and invitations that get "ignored, left pending, or marked as spam".

What LinkedIn does publish is durations, not thresholds. It says "Most restrictions will automatically be removed within one week", and that after withdrawing an invitation you cannot resend to that person "for up to three weeks". Those are recovery facts, and they are the only hard figures on the page.

This matters more than it looks. If you have been planning against 100 a week because a comparison article stated it, you have been planning against a number LinkedIn never published. The honest position is that the ceiling is unpublished and behavioural, which is exactly why a fixed number typed into a settings field is a poor safety model.

It is worth separating two things people conflate: invitations you send, and invitations sitting unanswered. Neither has a published limit, and the second is still a signal, which is why clearing stale pending invitations is hygiene regardless of volume.

Why the advice online disagrees with itself

Because each answer comes from a different account. A five year old profile with thousands of connections and daily activity is a different risk object from a seat created last month, and both authors may be reporting honestly about the account in front of them.

There is also a survivorship effect in the published numbers. The operator who ran hard and kept the account writes the post. The one who lost the account writes nothing, or writes a different post that never mentions the daily figure.

The practical response is to stop hunting for the real number and treat any figure as conditional on the account it came from. The question that matters is not what is allowed. It is what your seat can send this week while keeping acceptance inside the healthy band.

The number we run, and why it sits under the ceiling

A new seat here is capped at 20 invitations a day and 80 a week, ramped to 0.5x in week 0 and 0.75x in week 1. Since LinkedIn publishes no ceiling, we are not sitting under a documented one. We picked a number we can defend from our own sending history and then made it un-raisable.

The weekly cap binds before the daily one. Twenty a day over five working days would be 100, so the week runs out first. That spreads the sending instead of letting two hard days consume the allowance, which is also how a person working a list actually behaves.

The caps are enforced server side, not typed into a settings field. That distinction is the whole point. A number you can raise in a dropdown is a suggestion, and the request to raise it always arrives on the week a target is due.

Acceptance rate is the real limiter

Volume in isolation is not the thing that gets scored. Volume that continues while acceptance falls is. Under 20% acceptance means targeting is broken. 20 to 30% means relevance is thin. 30 to 50% is healthy, and our own 41% average across 861 conversations sits there. 50 to 60% is a strong signal match. Over 60% usually means a warm audience rather than a cold motion.

Which gives a rule that beats any daily number. If acceptance is under 20%, your safe daily limit is zero until the list changes. If you are in the healthy band, the published caps are already conservative and there is no reason to push them.

This is also why two accounts can run the same daily figure and end up in different places. One writes to people who recognise the reason and lands most of its invitations, producing almost no negative signal. The other produces the same volume as a stream of ignored requests, which is a different object from the platform's side.

What to do when you need more volume

Three real options. Tighten the list so more of the same volume lands. Add seats, which is what the Operator plan at $59 a month for up to three accounts and the Agency plan at $119 for up to ten exist for. Or accept a slower calendar than the one in the spreadsheet.

The option that is not on the list is raising the per seat number. Buying capacity by pushing one account harder is how the account becomes the constraint, and a restricted seat costs more working days than the extra volume was ever worth.

There is a timing lever teams forget as well. Sending inside working hours in the recipient's timezone, with real jitter between sends, costs nothing in volume and removes one of the easiest patterns to spot. It is not a substitute for any of the three options above, and it is free.

What the limits are not

They are not a promise of safety. Nothing at any pace is a promise, which is why we measure restrictions per managed seat per rolling 90 days and will publish the number when the denominator supports it rather than quoting one now.

They are also not the whole risk surface. Working hour pacing with real jitter, copy that varies between recipients, suppression of anyone already contacted, and an approval gate before every send all matter. A dedicated IP sits well below any of them, despite being the thing most vendors lead with.

They are not quality control either. Nothing in a cap prevents you sending 20 bad invitations a day for a month. The cap limits the rate of damage. The list decides whether there is any.