How we keep your LinkedIn account safe

LinkedIn account safety is a feature, not a marketing section. The accounts running outbound on Funkel are the same accounts you use to talk to your customers, your co-founders, and the recruiter who wants to ask about that job you held in 2019. A flag or a temporary restriction does not just kill your outbound: it removes you from your own professional network for as long as it takes to appeal.

We take this seriously enough that the safety logic in Funkel is more code than the message-generation logic. The pieces below are the actual rules that ship in the product.

The default caps

Each connected sender starts with 80 invites per week and 80 messages per week. LinkedIn’s absolute soft cap is 100 invites and 150 messages per week. We chose defaults that sit comfortably under that ceiling because the cost of pushing the limit and getting flagged is much higher than the cost of sending a few fewer messages this week.

The weekly cap is what the algorithm sees. We turn it into a daily budget by dividing across the campaign’s active days. Per default, that is five business days, so ~16 invites and ~16 messages a day per sender. Hit the daily ceiling and the queue waits until tomorrow. We never cram a day to clear a backlog.

The 14-day warmup

The most underrated thing in the safety stack. When you connect a sender, Funkel reads the connection date and treats the account as “warming up” for the first fourteen days. During warmup, the daily budget is scaled by a factor that ramps from 20% on day zero to 100% on day fourteen.

The math, in case you want it: factor = 0.20 + (daysSinceConnection × 0.06). Day 5 is 50%. Day 10 is 80%. Day 14 is 100%. After that, full caps.

We added this because new accounts that immediately run 16 invites a day are the ones LinkedIn flags first. A warmed account at 16 invites a day looks normal. A seven-day-old account at 16 invites a day does not.

Jitter between sends

Two messages going out 0.4 seconds apart from the same account is the signature of an automated tool. We never do that. Every send is delayed by a random interval between 30 and 180 seconds, and during the warmup period that window widens by an extra 0 to 120 seconds depending on how new the account is.

Sends also respect business hours. Each campaign has a schedule (days of the week, start hour, end hour, in your timezone); outside that window, actions sit in the queue and pick up at the next start time.

Reinvite cooloff

If a connection invite goes 14 days without a response, Funkel withdraws it. After a withdrawal, the same lead is locked out of new invites for 21 days. LinkedIn will flag accounts that re-invite the same person back-to-back, so we hold the line on the cooloff even when the agent finds a new signal for the same person inside that window.

What we pause automatically

• A reply lands. Every remaining workflow step for that lead is canceled in the same transaction.
• The weekly cap is hit. Sends pause; the queue resumes next week.
• A sender disconnects (token expired, LinkedIn invalidated the session, you signed out). Pending actions wait until you reconnect.
• The subscription enters the disabled state after a failed payment. Agents pause; senders stay connected but stop sending.

Retry and backoff

Transient send failures (a network blip, a LinkedIn API 500) get retried up to three times via the job queue’s exponential backoff. Permanent failures, like a reinvite cooloff or a deleted target profile, are caught at the first attempt and the action is marked skipped without retry. We do not hammer LinkedIn with retries on errors that would not resolve themselves.

What you should still do

Funkel can pace your sends but it cannot keep your account looking real. The accounts that survive long outbound runs are the ones that also post occasionally, comment on real conversations, and have a complete profile. Treat Funkel as one channel that runs alongside your normal use, not a replacement for it.

If LinkedIn ever shows you a warning or a temporary restriction, pause your campaigns immediately and let the account rest for at least seven days before resuming. The defaults above are tuned to keep you well clear of the line. They are not a guarantee. The only guarantee is the boring one: send fewer messages, send them with a reason, and stop the moment something looks off.