Monitor and troubleshoot app performance
Performance Insights gives organization admins and Jira and Confluence site admins visibility into the load time of key experiences for their users. It shows load-time trends, identifies factors that affect performance on the client's side, and highlights content or boards that may need attention.
Performance Insights is available in beta on the Enterprise plan.
How performance is measured
All load time data in Performance Insights is based on a metric called Time to Visually Complete (TTVC). It measures how long it takes from when a user starts loading a page to when 90% or more of the visible viewport has finished rendering.
TTVC is the primary metric we use internally to monitor and optimize the performance of Atlassian apps. Unlike First Contentful Paint (FCP), which fires as soon as any text or image appears, or Largest Contentful Paint (LCP), which tracks only the single largest element, TTVC waits until nearly everything the user can see has rendered. TTVC reflects the full end-to-end experience, not just the first signal that something is loading.
Key concepts
Performance bands
Load times are collected for all activity by all users in a given site. What you see is real user monitoring, not tests or samples.
Load times are classified into three categories based on TTVC:
Category | Threshold | What it means |
|---|---|---|
Fast | Less than 3.7 seconds | Users experience fast and responsive loads |
Moderate | 3.7 to 5.9 seconds | Users experience noticeable but manageable loads |
Slow | Greater than 5.9 seconds | Users are likely experiencing frustration |
These categories are based on a combination of data, customer sentiment, and industry standards. These thresholds apply consistently across Jira and Confluence.
Percentiles: P50 and P75
Performance Insights reports two percentile values:
P50 (Median) represents the typical experience. Half of all load events were faster than this value, and half were slower.
P75 represents the user experience in the slower quarter. 75% of load events were faster than this value.
Both values are shown as trend lines on load time charts, so you can see how performance changes over time.
Time ranges and granularity
You can view data across several time ranges and with up to 15-minute granularity. All time ranges use rolling periods (not calendar-anchored) and display in your local timezone.
Data availability begins from September 2026. Data from previous months and years will not be available.
Before you begin
Your organization must be on an Enterprise plan.
You must be an organization admin or site admin to access Performance Insights.
Your site must be a production site (not a sandbox).
Find Performance Insights in Atlassian Administration
Organization admins can access performance insights in Atlassian Administration. You will see overall and app-level performance, aggregated across all Jira and Confluence sites using Enterprise plan. Other Atlassian apps are not available in Beta.
To access:
In admin.atlassian.com, go to Insights, then System Health.
Select Performance in the left sidebar.
Performance by app
Select Performance by app to see performance on an app-by-app, site-by-site basis. This single view captures all Enterprise plan sites and helps to you identify problematic sites. Admins can select any app/site to open up a sidebar with additional details about that app/site's performance and potential improvement options.
Find Performance Insights in Jira
Jira site admins can access performance insights within the app.
Select System settings in the navigation.
Select Performance in the left sidebar.
Experiences tracked in Jira
Performance Insights tracks load times for three Jira experiences:
Board load time (Scrum or Kanban board)
Work item load time
Backlog load time
Overview
The Overview tab is the default landing page for Performance in Jira. It provides a snapshot of your site's performance health across three areas.
Rovo
Admins can invoke a Rovo Skill to get performance recommendations specific to your Jira site. Rovo will analyze all relevant data and prioritize any actions for you to take.
Load times
Load times shows how fast each experience loads over time. For each experience, you'll see:
P50 and P75 trend lines showing how median and 75th-percentile load times change across the selected time range
Performance band chart (stacked area) showing the percentage of load events that fall into each band (Fast, Moderate, Slow) over time
Event volume is displayed as a companion chart, so you can see whether changes in load time coincide with changes in traffic
Use the time range selector and granularity controls at the top of the page to adjust the view.
User setup
User setup helps you determine whether client-side factors contribute to slower load times. It breaks down performance across four dimensions:
Dimension | Cohorts |
|---|---|
Network speed (downlink) | Less than 5 Mbps, 5–10 Mbps, 10+ Mbps |
Network protocol | HTTP/1.1, HTTP/2, HTTP/3 |
Hardware (CPU cores) | Less than 4, 4–7, 8+ |
Operating system | Mac, Windows, Linux, Other |
For each dimension, you'll see a summary tile with one of three states:
Performance impact: A meaningful performance gap exists between cohorts. At least 10% of traffic comes from the slower cohort, and the gap is 250 milliseconds or more.
Optimal setup: Performance is consistent across cohorts, with no significant gaps detected.
Inconclusive data: There isn't enough data to make a determination, or cohort behavior doesn't follow expected patterns.
Select any tile to see a detailed breakdown with P75 load times per cohort, percentage of events in each cohort, and trend charts.
Board performance
Board performance shows load times for individual Jira boards, broken down by recorded speed and usage volume.
The table includes columns for board name, performance status, load time, and number of views. You can sort by any column and use the search bar to filter by board name or board admin.
Find Performance Insights in Confluence
Confluence site admins can access performance insights within the app.
In the Confluence site administration, go to Monitoring and select Performance.
Experiences tracked in Confluence
Performance Insights tracks load times for three Confluence experiences:
Page View load time (published Confluence pages)
Page Edit load time (opening the page editor)
Live Docs load time
Load times
Load times shows how fast each experience loads over time. For each experience, you'll see:
P50 and P75 trend lines showing how median and 75th-percentile load times change across the selected time range
Performance band chart (stacked area) showing the percentage of load events that fall into each band (Fast, Moderate, Slow) over time
Event volume is displayed as a companion chart, so you can see whether changes in load time coincide with changes in traffic
Use the time range selector and granularity controls at the top of the page to adjust the view.
Content metrics
Content metrics identifies individual pages, live docs, and blog posts with slower load times. This helps you pinpoint specific content that may benefit from optimization (For example, pages with heavy embedded content, large tables, or many macros).
At the top of the tab, a summary bar shows the overall distribution of your content across performance bands (Fast, Moderate, Slow). Below that, a table lists individual content items with their load time rating, median load time, view count, space, and content owner.
Content from spaces you don't have access to appears as "Restricted" with a lock icon. You can still see the load time data, but the content title and space name are hidden.
Recommended actions
Spot trends in Load times
Look for sustained increases in P50 or P75, or a growing share of Slow events in the performance band chart. Short spikes are normal; multi-day or multi-week trends warrant investigation.
Check whether the cause is on the client’s side
For Jira, use User setup in Performance insights to see whether slow loads correlate with specific network speeds, protocols, hardware, or operating systems. If a particular cohort is driving the slowdown, the issue may be on the user's end rather than in the app.
Identify specific problem areas
Use Board performance to find individual boards with slow load times in Jira, or Content metrics to find pages, live docs, or blog posts that load slowly in Confluence.
Take action
For slow Jira boards, reach out to the board admin to review board configuration. Complex JQL filters are a common cause of slow boards, and Rovo can help optimize JQL queries. For slow Confluence content, work with the content owner to reduce embedded content, large tables, or macro-heavy layouts.
Was this helpful?