Why LinkedIn restricted your outbound account (and how the next one comes faster)

Your account just came off restriction. You think you got a slap on the wrist. You did not. You got a flag on a permanent record, and the model is now watching every action you take from this point forward at a tighter threshold than it watched yesterday.

Here is the part nobody tells you when you read the LinkedIn help center: the second restriction comes faster than the first. The third is often permanent. The interval between events shrinks each time because LinkedIn weights account-level risk history heavily in its real-time decision on every single action you take.

If you came here from a Google search the day after a 24-hour suspension lifted, the most expensive thing you can do today is start sending again to "make up the volume". Every accepted request right now is one the algorithm is watching for a reason to escalate.

The 4 signals that trip the model

The risk model is not a single rule. It is a composite watching for the unhealthy combination of:

1. Volume relative to your account's historical baseline. A 5-year-old account that suddenly starts sending 40 requests/day after years of 5/day raises a flag faster than a 1-year-old account doing the same thing from day one. The model knows what "normal for you" looks like.

1. Acceptance rate. Sub-15% acceptance pulled across 50+ requests is the cleanest "unwanted contact" signal LinkedIn has. The model does not need your prospects to report you. It can see that they are ignoring you.

1. IP fingerprint. Same IP for every action across thousands of requests is the easiest bot signal there is. Residential rotation is the table-stakes fix. Most operators run an outreach tool that handles this automatically; if you do not, every single action you take leaks the same identifier.

1. Action variety. Healthy human accounts do many things: read articles, like posts, comment, search, send messages, accept requests. An account that only sends connection requests + DMs looks like an outbound tool, regardless of how the requests are sent.

The 14-day plan

You have one window. The second restriction will come within 30 days unless you change the inputs. Here is the cheapest possible recovery:

Days 1-3: Stop ALL outbound from the restricted account. Not "reduce volume". Stop. Use the time to read 4-5 posts from accounts in your target ICP and comment on them; this is the only outbound action the algorithm reads as positive at this stage.

Days 4-7: Send 3 connection requests per day. Each one to someone whose post or profile you just engaged with. Acceptance should run 60%+ at this volume because the requests now have context. Continue commenting on other accounts.

Days 8-14: Scale to 8 requests per day, holding acceptance above 35%. Add one DM per day to recently-accepted connections, opening with one specific question.

By day 14 the account is back to healthy and the risk-model state has decayed by roughly 60% of its post-restriction peak. You can continue scaling from there at the rate acceptance allows, capping at 20-25/day per account.

What you should do BEFORE the next restriction

If you only have one outbound account today and your livelihood depends on it, the next restriction is a pipeline event, not an account event. The fix is structural:

• Provision a second account NOW. 4-6 weeks of warmup is required. Doing it after the next restriction means a month of zero pipeline.
• Move outbound volume across a pool of 4-6 aged accounts so no single one bears visible volume. This is what every team that runs outbound at scale ends up doing, and it is the single biggest restriction-rate reduction available.
• Audit the trigger that caused the first restriction. Was it volume? Acceptance? IP? Templates? The model writes a reason; figure out which signal caused yours so you do not retrigger the same one from the new account.

Want to know how close your current setup is to the next restriction? Take the 60-second BIGM diagnostic. 8 questions, 0-100 score, ban-risk band, the exact gaps that are tipping the algorithm against you. No signup required to see the score.