Each monitor keeps its own uptime percentage, response-time chart and event log. Every outage is recorded with its cause and how long it lasted, so a conversation about reliability starts with data rather than memory.
What the failing check actually reported, whether that was a refused connection, a timeout, a wrong status code or a missing string in the body.
When the monitor went down and when it recovered, so the length of an incident is a fact rather than an estimate.
A running figure per monitor, which is the number people ask for when they want to know how reliable something has been.
How long checks have been taking over time. A service getting slower is usually visible well before it fails.
The sequence of state changes for a monitor, so you can see the pattern rather than a single incident in isolation.
A log step writes a note to the timeline without alerting, which is a clean way to mark escalation levels in the record.
Start from what actually happened and when, rather than from three people's recollection of a bad afternoon.
Response-time trends catch the slow decline that never quite trips an alert until it does.
Per-monitor uptime gives you a defensible number when someone asks how a service has been performing.
An optional Monday recap of how your monitors did the week before. It is off by default, so it is there only if you want it.
Three monitors are free forever. Start recording cause and duration on everything you run.