Outlier Incident
View the frequency of similar incidents to gain helpful context during incident triage
The Outlier Incident feature provides incident responders with immediate context on how often similar incidents have occurred on a specific service, helping them quickly determine whether an issue is familiar or new. This is particularly valuable for IT operations teams and DevOps engineers who need to identify patterns and irregularities to make informed decisions during response efforts. By categorizing incidents as frequent, rare, or anomalous, the feature streamlines the analysis process, allowing teams to focus on resolution and ultimately improve service reliability.
Availability
This feature is available with our PagerDuty AIOps add-on, or with Legacy Event Intelligence. If you would like to sign up for a trial of PagerDuty AIOps features, please read PagerDuty AIOps Trials.
View Outlier Incident in the Web App
- From the Incidents page or a service’s detail page, click your desired incident’s Title to see its detail page.
- If an incident has been categorized as a frequent, rare or anomalous incident these details will appear below the incident title.
![Incident with "Anomaly" Outlier Incident label A screenshot of the PagerDuty web app UI indicating an incident with the "Anomaly" Outlier Incident label](https://files.readme.io/125a679-4d036c6-outlier-incident-overview.png)
Incident with "Anomaly" Outlier Incident label
Outlier Incident Labels
Outlier Incident uses data science techniques to dynamically map incidents to templates and determine how often each type of incident occurs on a service. The following labels are applied below the incident title on the incident's details page:
- Frequent: Incident types representing 20% or more of all incidents in the previous 30 days.
- Rare: Incident types representing 5% or less of all incidents in the previous 30 days.
- Anomaly: Incident types that have not occurred on the service in the previous 30 days.
For the purpose of determining which of the labels above to use, incident data from the 30 days prior to an incident’s trigger date is used.
Updated 9 days ago