How to spot invites sent outside your tool before they wreck your limits

Compare the pending invitations on the LinkedIn account with the leads your tool invited. Any pending invitation that matches no lead was sent from somewhere else: by hand, from another tool or from a second user of the same profile. Because manual and agent invitations share one weekly limit, a few outside invites can push an account past what your tool believes it has used.

This is a small problem that causes large surprises. An account looks healthy in your outreach tool, with 90 invitations out of a 150 weekly cap, and then LinkedIn says the limit is reached. The missing 60 were sent by a colleague on their phone, or by an old extension nobody switched off. This post explains how to find them and what we built to do it for you.

Why does it matter?

LinkedIn counts invitations per account, not per tool. Your outreach platform keeps its own count of what it sent. When invitations also come from the app, a browser extension or a second tool, the platform's count is lower than LinkedIn's. The tool's headroom is an illusion.

The result is the failure you wanted to avoid: reaching the limit sooner than planned and drawing attention. LinkedIn lists ignored and pending invitations among the reasons for a restriction (types of restrictions), and manual sends often go to people you know less well. See LinkedIn invitation limits for what LinkedIn says and does not say about numbers.

Where do outside invites come from?

• You or a teammate, by hand. The most common. Sending a few invites from the app after reading a profile.
• Another automation tool. An old tool left connected.
• A browser extension. Often forgotten.
• A second workspace. The same LinkedIn profile connected to two accounts in the same platform, or in two tools.
• A recruiter or sales seat. Some seats send from the same profile through another product.

How can you check by hand?

1. Open the account's sent invitations on LinkedIn. It lists pending invitations with the date they were sent.
2. Export or list the people your tool says it invited in the same period.
3. Match them. Anyone pending who is not in your tool's list was sent elsewhere.
4. Count the last seven days. Add the tool's count and the outside count, and compare with your weekly cap.

Two cautions. Dates on pending invitations are day accurate, but the time of day is not reliable in what the platform returns, so do not try to match by minute. And an invitation that was accepted leaves the pending list, so what you find is a floor.

What does Sellinger do?

Sellinger runs this comparison for you. About once a day per seat, only while the agent has new leads to invite, it takes the pending invitations from the last seven days that were sent after the account was connected, and checks whether each one matches a lead. If an invitation matches no lead, it counts as outside the tool.

A warning shows when three or more outside invitations appear in the last seven days, the agent is also inviting from the seat, and the check is recent. You see it in three places: an amber icon on the LinkedIn account card, a row in Account health on the dashboard, and the notification bell.

The matching trap

The first version of this check gave false alarms, and the reason is worth knowing if you build your own. One LinkedIn profile is often connected under several workspaces: a person and their assistant, an agency and a client, a test workspace. Matching pending invitations only against the leads of one workspace made invitations sent by the other workspaces look manual. In our first dry run across production this produced 70 false manual invites. The check now compares against the leads of every workspace connected to the same LinkedIn profile.

The lesson applies to any tool: before you call an invitation manual, make sure no other user of the profile sent it.

What should you do with a warning?

1. Find out who sent them. Ask the team. Look at the dates and the people.
2. Stop the source. Remove the extension, disconnect the old tool, or agree that invites come from one place.
3. Count them against the cap. If outside invites took 40 of the 150, lower the agent's weekly cap for the week.
4. Check acceptance. Manual invites to weak contacts can lower acceptance. Watch it, since about 4 in 10 invitations were accepted across 42 customer teams in the 90 days to 19 September 2026.
5. Keep a rule. One place sends invitations from an account. Everything else is replies and messages.

What it cannot see

• Invitations that were accepted or withdrawn before the check.
• Messages and other activity, which have their own limits.
• Activity on a different account or profile.

So treat the warning as a smoke alarm. It does not prove a problem, and its silence does not prove there is none.

What does a made up week look like?

An illustration with invented numbers. An account has a weekly cap of 150. The agent sent 90 invitations from Monday to Thursday. On Friday the owner opens the account's sent invitations and finds 148 pending from the last seven days. Matching them against the tool's list leaves 58 that match no lead.

• The tool believes it has 60 invitations of headroom.
• LinkedIn sees almost no headroom, because 58 more came from elsewhere.

The owner finds out that a teammate sent invitations from the phone app after a conference and a forgotten extension sent the rest. They pause the agent for the week, disconnect the extension and agree that invitations come from one place. The weekly cap does not move, but the plan around it does.

How do you reduce the chance of outside invites?

• Say it out loud. One sentence in the team handbook: invitations from this account come from the tool.
• List every tool connected to each profile. Include extensions, other platforms and recruiter seats.
• Ask at onboarding. When an agency takes over a client's account, ask which tools have ever touched it.
• Review monthly. Open the sent invitations page and compare it with the tool's list once a month, even when no warning has appeared.

What if the warning is wrong?

It can be, and the first version of our check was. The causes to rule out, in order:

1. Another workspace on the same profile. Check whether the same LinkedIn profile is connected under a second workspace, an assistant's login or an agency account.
2. An invitation from before the connection. The check looks at invitations sent after the account was connected, but a mismatch in dates is worth a second look.
3. An old campaign. Leads that were removed from a list may still have a pending invitation.

If none of these explains the invitations, treat them as real.

Does it matter if the outside invites are few?

Three or more in a week is where the warning appears, and the reason for the threshold is practical: one or two can be a mistake and a larger number changes your headroom. The cost of a few is small if you count them against the cap. The cost of ignoring a pattern is that your tool's count and LinkedIn's drift apart week after week. Keeping them close is the whole point of the routine above.

Where do you see the warning?

In three places in Sellinger: an amber icon on the LinkedIn account card, a row in Account health on the dashboard and the notification bell. The steps for limits and pacing are in daily limits and pacing.

A short routine

• [ ] One tool sends invitations from each LinkedIn account.
• [ ] Extensions and old tools are disconnected.
• [ ] Anyone who needs to invite someone tells the owner, and the week's cap is adjusted.
• [ ] I check the account card and Account health weekly.
• [ ] I know my weekly cap and what share is used by the agent.

More on keeping an account healthy is in the 15 point LinkedIn safety checklist.