Automation Engine
API & Webhooks
Past incidents
How we measure
Only failures on our side count. An action fails when our platform could not complete it – an internal error, a crashed or interrupted run. These are not failures: LinkedIn's legitimate answers (person not found, already connected, LinkedIn's own limits), limits you configured, invalid input, and your account's own state (signed out, unsupported interface language).
Status follows breadth, not volume. Every action group is evaluated over a rolling 3-hour window with at least 10 actions:
- Degraded – 5% or more of actions fail on our side, across at least 3 accounts
- Partial outage – 15% or more, across at least 5 accounts
- Major outage – 50% or more across at least 5 accounts, or the API is unreachable
Every partial or major outage gets an incident report. A problem limited to a single account – its LinkedIn session or proxy – never changes this page; you will see it in your dashboard.
Start time is measured from the moment your account is free to the moment the action begins, so your own queue of earlier actions is not counted. Start and duration times are estimates from our activity logs.
API history before we began measuring requests is reconstructed from the requests we accepted, hour by hour. Those days are marked in their tooltips and count toward the 90-day uptime.