How do I stay connected on LinkedIn and stop my session from disconnecting?
π€ What's actually happening when you get disconnected?
When you log into LinkedIn, it creates a session - essentially a record that you're logged in. PhantomBuster uses that session to act on your behalf. When LinkedIn ends that session, your Phantoms stop working and you see a disconnection error.
You might have heard this called a "session cookie" issue. The cookie is just how that session gets stored - but the part that matters is simpler: LinkedIn decided your session was over, and PhantomBuster lost access.
The important thing to understand here:
That distinction shapes everything in this guide.
β Why does it happen?
Based on our data, two patterns cause the majority of disconnections.
π Pattern 1 - The Stop and Start (most common)
This shows up in roughly 65β74% of disconnections. The pattern looks like this:
- You've been running automation regularly
- Then you go quiet for a while, maybe a holiday, a busy week, or just forgetting to run your Phantoms
- Then you come back and fire everything up again
What makes this risky isn't the restart. It's the gap before it. LinkedIn's systems look for consistent, predictable behaviour. When an account that's been silent for a week suddenly starts firing off actions, that inconsistency triggers a flag - even if the volume is the same as before.
The gap is the problem. Not the spike.
π Pattern 2 - The Cold Start (new accounts)
LinkedIn builds a behavioural profile for every account over time. If your account has never sent 30 connection requests in a week and you suddenly start doing it every day, there's no history to compare it against, it just looks like an automated account appeared from nowhere.
New accounts and reactivated accounts are at higher risk in the first couple of weeks. This doesn't mean you can't automate, it means you need to build a baseline first.
π² What about the other ~34%?
About a third of disconnections don't fit either pattern. LinkedIn runs platform-wide enforcement periodically, and it can catch accounts doing everything right. Factors outside your control (IP address, device, account age) can also play a role.
No guide will make you completely immune. What we can do is give you the best possible chance of staying connected.
β How to prevent disconnections
1οΈβ£ Don't go dark
This is the single most important thing. The biggest risk factor is letting your account go quiet for more than 3β4 days and then jumping back to full speed.
If you can keep some level of activity running, even light, even low volume, you're dramatically reducing your disconnection risk. Even running a small scrape a few times a week is enough to keep your account's behavioural baseline alive.
2οΈβ£ Ramp up slowly after any break
If you've taken time off, don't restart at full speed. Here's a safe approach:
| Day | Max volume | Focus |
|---|---|---|
| Days 1β5 | ~20% of normal | Light activity - profile views, small scrapes |
| Days 5β10 | ~40β50% of normal | Start adding connections gradually |
| Days 11β15 | ~75% of normal | Messaging, follow-ups |
| Day 15+ | Full rate | Back to normal if everything's been stable |
3οΈβ£ Warm up new accounts
If you're starting PhantomBuster on a LinkedIn account that hasn't been automated before, follow this two-week warm-up:
| Period | Daily cap | Notes |
|---|---|---|
| Days 1β5 | 10 launches/day | Very light - just establishing a baseline |
| Days 5β10 | 25 launches/day | Start building the profile gradually |
| Days 11β15 | 50% of target rate | Watch for any warnings |
| Day 15+ | Full rate | Fine to push if nothing's been flagged |
The sweet spot for ongoing operation is 5β15 launches per day. Users in that range survive at a significantly higher rate than those running under 5 or over 30.
4οΈβ£ Keep your timing varied
LinkedIn looks at how you act, not just how much. Actions at perfectly regular intervals look machine-like. Real people are inconsistent.
In PhantomBuster, use delay ranges rather than fixed values. A 30β90 second range looks far more natural than a fixed 30-second delay.
- Keep minimum delays above 20β30 seconds
- Don't run automation for more than 6β8 hours per day
- Stay under 15 combined actions per hour across all Phantoms
5οΈβ£ Watch for warning signs
LinkedIn will usually give you signals before a full disconnection. Take these seriously:
- More errors than usual in your Phantom runs
- CAPTCHA challenges when logging into LinkedIn manually
- LinkedIn asking you to verify via email or phone
- Connection acceptance rates dropping noticeably
π¨ What to do when you get disconnected
It happens to everyone. Here's the step-by-step:
- Stop everything immediately - don't retry, don't refresh, don't run Phantoms
- Wait - at least a few hours, ideally until the next day
- Log out of LinkedIn and back in manually - this gives you the freshest possible session before reconnecting
- Reconnect in PhantomBuster - use the "Connect to LinkedIn" flow to grab the new session from your active browser. Confirm it shows as active before doing anything else
- Wait another few hours - give the new session some breathing room
- Come back gradually - start at 10-20% of your normal rate, and increase by no more than 10β20% every few days
- Only return to full speed after a stable week
Once you're back up, take a moment to review what happened. Had you taken a break before the disconnection? Had your volume spiked? If you're getting disconnected more than once a month on the same account, something structural needs to change.
π Quick Reference
Daily habits that protect your account
| Habit | Why it matters |
|---|---|
| Stay active at least every 3β4 days | Gaps are the #1 cause of disconnections |
| Run 5β15 launches/day as your baseline | This is the safe operating zone |
| Use variable delays (30β90 seconds) | Fixed intervals look machine-like |
| Cap active hours at 6β8 per day | Humans don't work LinkedIn around the clock |
| Ramp up after any break of 5+ days | Abrupt restarts look suspicious |
| Warm up new accounts over 2-4 weeks | Build a baseline before pushing volume |
Recovery checklist
- Stop all automation immediately
- Wait a few hours (ideally overnight)
- Log out of LinkedIn and back in manually
- Reconnect session in PhantomBuster
- Wait another few hours before running anything
- Restart at 10β20% of normal rate
- Increase 10β20% every few stable days
- Full speed only after a stable week
Comments
0 comments
Please sign in to leave a comment.