Monitoring observation limits
Find what samples, averages and probe paths can actually establish.
Find an outage hidden between polling instants
Compare an actual interval with point observations instead of inferring continuous health.
Distinguish consecutive bad samples from elapsed failure time
Apply a three-sample trigger without claiming three minutes of known failure.
Find a failing region behind a healthy-looking average
Compare weighted overall success with a separate regional acceptance rule.
Report missing monitor samples without counting them as success
Show observation coverage and observed success as two different ratios.
Check whether probes cover the whole tested journey
Compare observed path edges with a required journey rather than its landing screen.