GitLab CI Insights - New CI/CD Analytics view for projects (Limited Availability)

Note

The new CI/CD Analytics view depends on Clickhouse for the database layer. If you do not configure Clickhouse for your self-managed GitLab instance, then the legacy CI/CD Analytics view will be displayed instead of the new option.

Overview

The improved GitLab CI view provides developers with metrics and data visualizations in the GitLab UI for critical CI/CD pipeline performance and success rates. As these metrics are available within the project repository, development teams have immediate access to the data they need to understand at a glance the health and efficiency of the software development process for a specific project.

The new view includes the following metrics, charts and filters

CI/CD pipeline metrics:

Total pipeline runs
Median duration
Failure rate
Success rate

CI/CD trend charts:

50th & 95th percentile pipeline duration over time (week, last 30,90,180 days)
Pipeline status (successful, failed, other)

Report Filters:

Pipeline source
Branch

Problem to solve

The Project CI/CD analytics page contains a lot of data that is confusing for the user to understand and identify next steps with when evaluating the health of their pipeline. See more issues that were identified in gitlab-design#2439 (closed). Based on problem validation, we need to refactor the page to allow the user to ingest the data meaningfully and give them an easy way to identify pipeline health.

JTBD we are solving for: When I am managing continuous integration of code at scale, I want to understand the pipeline health, so I can successfully resolve and prevent issues from occurring.

Intended users

Software developers
Platform engineers
DevOps engineers

User experience goal

The user should be able to quickly understand how healthy their pipeline is, specifically focusing on status and duration.

Proposal

Acceptance criteria

Criteria	Current UI	With this proposal	Component needed	Minimal required database engine
Update single stats on the page			Single stat	Clickhouse
Move individual filters on graphs to be universal filters on the page. Add a new filter for `source`. Add a new filter for `branch`.			Combobox	Clickhouse
Update page title.	`CI/CD analytics`	`Pipelines`	N/A just text.	N/A
Update duration chart to be a line graph and to show p50 and p95. Use the color system for categorical data for colors on the graph.			Line chart with legend interaction	Clickhouse
Update status graph to include failures, successes, and other (cancelled + skipped) runs. Use custom colors for the graph, successful=`dv-green-500`, failed=`dv-magenta-600`, other=`dv-blue-500` .			Stacked column chart with legend interaction	Postgres
Resources 🔗

Available Tier

GitLab Free

Feature Development Stage

Limited availability

Are ready for production use at a reduced scale.
Available on one or more GitLab platforms: GitLab.com
Available on one or more GitLab platforms: GitLab Self-Managed (Certain metric charts depend on Clickhouse
Available on one or more GitLab platforms: GitLab Dedicated (Certain metric charts depend on Clickhouse
Are fully supported and documented.
Have a complete user experience aligned with GitLab design standards.

Feature Usage Metrics

What does success look like, and how can we measure that?

Measuring adoption:

we track visits to the page today, so we'd want to see an increase.

Measuring if the changes build a better experience:

Through a feedback banner (tracked in a separate issue). This could also help us figure out which metrics are most important to add next.

Proposed API layout

{
  project(fullPath: "rf2-top-level-group-1/rf2-group-1.1/rf2-group-1.1.1/rf2-project-1-1-1-1") {
    pipelineAnalytics(finishedAfter: "2023-03-28", finishedBefore: "2024-03-28", source: PIPELINE, branch: DEFAULT) {
      # Single stat
      aggregate {
        # label # Not needed here, would contain formatted date range
        count
        successCount: count(status: SUCCESS) # Success rate can be computed from successCount / count
        failedCount: count(status: FAILED) # Failure rate can be computed from failedCount / count
        durationStatistics {
          p50
        }
      }

      # Pipeline charts
      timeSeries(period: MONTH) {
        label # Timestamp of beginning of period

        count
        successCount: count(status: SUCCESS)
        failedCount: count(status: FAILED)
        durationStatistics {
          p50
          p95
        }
      }
    }
  }
}

Changes required in backend

GraphQL: Refactor `PipelineAnalytics` and `Pipe... (#479285 - closed)
1. Add a total field of type PipelineAnalyticsPeriod to PipelineAnalytics
2. Add alpha time filters to PipelineAnalytics type
3. Change PipelineAnalyticsPeriod type to have label and a count field
GraphQL: Add support to compute mean duration o... (#454310 - closed)
1. Rename CiJobsDurationStatistics to CiDurationStatistics for reuse
2. Add durationStatistics field to PipelineAnalyticsPeriod type

This page may contain information related to upcoming products, features and functionality. It is important to note that the information presented is for informational purposes only, so please do not rely on the information for purchasing or planning purposes. Just like with all projects, the items mentioned on the page are subject to change or delay, and the development, release, and timing of any products, features, or functionality remain at the sole discretion of GitLab Inc.

Edited May 21, 2025 by Darren Eastman