LogRocket uses machine learning and LLMs (large language models) to recommend certain issues that your team may consider to be severe. Our issue recommendation models are trained on data from LogRocket sessions and issues, and we use them to generate severity scores for your issues.
Severe Issues are not available for self-hosted installations
Issues Severity Scores and Issues Digest are not yet available for self-hosted installations of LogRocket. For any further questions, contact [email protected].
Issues with higher severity scores are more likely to be associated with user frustration and may indicate that users were prevented from completing an in-app task. Issue severity scores are computed over a time range (e.g., last day, last week) and may change as you select different time ranges. An issue's severity score is also more accurate when more data is available for it. If you see a message that LogRocket doesn't have enough data to assign a severity score to an issue, try increasing the time range or wait for more data to be collected.
By default, LogRocket shows all issues sorted by severe first. To show only the severe issues, change the Severity filter from "Any Severity" to "Severe".
An issue's detail view will also display severity scores. The score indicator in header will show the issue's score as computed over the last 30 days:
The event list in the detail view will also display a severity score indicating the degree of frustration associated with the specific event listed. Events may be sorted by severity score or time of occurrence in descending order.
In the session replay view, if an event in the recording is associated with an issue, a severity indicator and a link to the issue will be shown alongside the event. In this context, the severity indicator displays the degree of frustration associated with that event.
Network Error issue events will appear in the Network pane:
Network traffic is sampled before being treated as an Issue
LogRocket samples network events before processing them as issues. Not all network errors will have an associated link to an issue and a severity score.
All other issue types (Rage Clicks, Dead Clicks, Error States, Frustrating Network Requests) will appear in the Event Timeline:
New sessions may not be processed immediately
It can take about an hour for severity scores to be populated for new sessions. If you see an empty severity score indicator, the corresponding session is likely still being processed.
LogRocket delivers a weekly Issues Digest to your channel of choice - Email, Slack, Teams, or Webhook. See how to Issues Digests work and are driven by Galileo ML.
LogRocket's issue recommendations are constantly improving, and you can help! Use our issue triage features to teach LogRocket that an issue is or is not severe.
For example, if you agree that a severe issue is severe, triaging it as "High" or "Low" impact will make those kinds of recommendations more likely in the future. If you don't think an issue should have been considered severe, triaging it as "Ignored" will make those kinds of recommendations less likely in the future. Similarly, if you think LogRocket should have recommended an issue but it didn't, triaging it as "High" or "Low" impact will make it more likely that similar issues are recommended in the future. And if you agree that LogRocket correctly decided not to recommend an issue, triaging it as "Ignored" will help to reinforce that behavior.
Changing issue group conditions may change severity score
Keep in mind that customizing grouping conditions may cause the triaged issue's severity score to change.
Updated 2 days ago