Your LinkedIn account is the asset. Every engine decision in NudgeLink is built around one rule: behave like a careful human, never like a bot. This article is the concrete list of what that means in practice.
Warmup before outreach — per lead
A sequence doesn't open with a connection request. For each lead the engine first behaves the way a genuinely interested person would:
- View the profile.
- Like a recent post (when there is one).
- Comment on a post — written about that specific post, and shown to you for approval like any other message.
- Only then send the connection request.
By the time your invite arrives, the lead has seen your name before — accept rates are higher, and the activity pattern reads as human because it is the human pattern.
Warmup ramp — per account
A freshly connected seat starts at low daily limits that increase week by week. New accounts doing too much too fast is the classic restriction trigger; the ramp makes your first weeks deliberately conservative. Invite volume ramps the same way.
Daily caps — always
Every action type — profile views, likes, comments, invites, messages — has a per-day, per-account cap that the engine enforces before acting, not after. Caps are tracked per calendar day in your account's own timezone. LinkedIn Premium seats get slightly higher discovery allowances; invite and message caps never depend on it.
Human send windows
Actions run only between 08:00 and 20:00 in the account's local timezone, at jittered moments — not on the hour, not in bursts, and not at 3am. Deferred actions spread out rather than firing all at once.
The approval gate
In the default review mode, no outreach text leaves the building without your explicit approval — and drafts wait for you indefinitely rather than expiring. A message that can't be reviewed can't be sent; a draft that fails our own deliverability checks is blocked before it ever reaches your queue.
Continuous account health monitoring
The engine watches each seat's health: connection status, LinkedIn warnings, and error patterns. If a seat looks unhealthy, its sequences pause automatically and resume only when the account recovers — the engine never pushes through a warning. If you disconnect a seat, the hosted session is severed at the provider; nothing keeps acting on your behalf.
What we ask of you
Two things: don't run other automation tools on the same LinkedIn account in parallel (stacked tools defeat every cap above), and keep your account's timezone accurate so the send window matches your actual working hours.