Monitoring is only useful if the right person hears about a problem while there is still time to act. Each reading and event is checked against the thresholds set for that sensor as it arrives, and an alert goes out straight away by email, SMS or both, or through an app where the client wants one. Thresholds are set sensor by sensor: a high or low limit, a reading that drifts slowly over several hours, a door left open outside agreed hours, a contact that changes state. On a 4G gateway, an alert reaches the dashboard about a second after the reading arrives.
Alerts go to named people for each site, with escalation to further contacts when the first person does not respond, so an out-of-hours alarm does not sit unread in one inbox. Silence counts as a fault as well. Every tag checks in on a schedule set for the site, so a tag that stops reporting raises its own alert, and gateways carry battery backup, with the platform alerted the moment one loses mains power. Every alert is kept in a timestamped event history, held in the cloud or on the client's own servers, and alerts are never metered or charged per message.