## Latest update : [MVC Prototype](https://group-ci-cd-analytics-dashboard-prototype-936033.gitlab.io/) This Prototype and its Dummy Data: Hierarchy Map are directly connected. Selections you make in either view carry across when you switch tabs. You can explore the full data structure on the Hierarchy Map, pick specific projects or scenarios, and then switch to the Prototype to see the dashboard populated with exactly that data. The chart now includes a previous period comparison line, the drill-down table supports Projects, Pipelines, and Jobs views, and new scenario buttons let you quickly surface patterns like shared pipeline or job names across projects. {width="270" height="56"} {width="256" height="56"} ## Problems to Solve Platform engineering teams and DevOps administrators managing CI/CD pipelines across multiple projects lack visibility into aggregate pipeline performance and health at the group/organization level. Currently, they must: * Navigate to individual projects to view pipeline metrics * Use external tools (Datadog, Grafana) to aggregate data across projects * Manually consolidate information when infrastructure changes impact pipelines * Spend significant time troubleshooting cross-project issues without centralized insights This fragmented approach delays identification of systemic issues, makes it difficult to assess organizational CI/CD health, and increases operational overhead. ## Customer Impact From our research with 46 survey respondents and [5 in-depth interviews](https://gitlab.dovetail.com/projects/1Mzvx28zaRTLqTJ4ENFJ7Z/v/406zFLjIKreaT8KcKWwhue): * Primary users: [Platform Engineer](https://handbook.gitlab.com/handbook/product/personas/#priyanka-platform-engineer), SREs, DevOps teams managing 7-100+ projects * Key pain points: * "We don't have a single pane of glass to check all pipelines" (Participant 2) * "It would be nice to have some quick access if we make a change...to see if this change has negatively impacted things significantly" (Participant 1) * "Getting everything in one place - data's all over, metrics differ, and it's hard to compare" (Survey respondent) ### ## Key Use Cases from Research <table> <tr> <th> #### Use Case 1: Infrastructure Change Impact Assessment </th> <th> #### Use Case 2: Proactive Morning Health Checks </th> </tr> <tr> <td> Scenario: Platform team changes runner node types or upgrades GitLab Runner Current Pain (Participant 1): * Made an infrastructure change to Kubernetes nodes * Took 2 weeks to discover pipeline performance degraded * Had to build custom GraphQL queries to analyze impact * Used Datadog to correlate timing with infrastructure change Desired Experience: * View group-level pipeline duration trends before/after change * Quickly identify if specific jobs are affected across all pipelines * Filter by time range to pinpoint exact moment of degradation </td> <td> Scenario: Daily standup preparation and health monitoring Current Pain (Participant 3): * Checks multiple systems (GitLab + Grafana) each morning * No quick way to see if yesterday's changes caused issues Desired Experience: * Dashboard showing overnight pipeline activity * Spike detection for failure rates * Trend comparison (today vs. 7-day average) </td> </tr> </table> ### Proposed Solution https://gitlab.com/gitlab-org/gitlab/-/work_items/519627#ux Add group-level CI/CD performance metrics to provide platform teams with: Location: Group Level -\> Analyze -\> New Dashboard -\> CI/CD Analytics 1. Show aggregated pipeline health and reliability indicators across all group projects 1. Pipeline success rate ("Pipeline success rate provides an immediate snapshot of how consistently pipelines are running across the organization" (Survey respondent)) 2. Job Failure Rate ("Job failure combines multiple projects and provides overall outlook for quick analysis" (Participant 3)") 3. p50/p95 Pipeline Duration 4. Total Pipeline Runs 2. Trend analysis to identify degradation over time 1. Pipeline Duration 2. Pipeline Success Rate 3. Job Failure Rate 3. Quick identification of systemic issues affecting multiple pipelines 1. Provide a way to spot which pipelines or jobs had a trend shift 4. Ability to filter and drill down to specific subgroups 1. Filter by subgroups 2. Time ranges: What default and custom time ranges should we support? * Common needs: Last 24 hours, Last 7 days, Last 30 days, Custom range * Infrastructure change correlation requires hour-level granularity ## Success Metrics | Metric | Target (6-month post-launch) | Measured By | Target Visualization | |--------|------------------------------|-------------|----------------------| | Feature Adoption | 25% of groups with 10+ active CI/CD projects | View the group-level CI/CD dashboard at least once | Tableau | | Feature Usage | 3 | Median views per active user per week | Tableau | ## Business Outcomes - Qualitative * Reduce external tool dependencies (Datadog, Grafana) for CI/CD monitoring * Increase Ultimate/Premium tier value ## Out of Scope * Usecase: Runner cost and optimization ## UX ## Latest : [MVC Prototype](https://group-ci-cd-analytics-dashboard-prototype-936033.gitlab.io/) This Prototype and its Dummy Data: Hierarchy Map are directly connected. Selections you make in either view carry across when you switch tabs. You can explore the full data structure on the Hierarchy Map, pick specific projects or scenarios, and then switch to the Prototype to see the dashboard populated with exactly that data. The chart now includes a previous period comparison line, the drill-down table supports Projects, Pipelines, and Jobs views, and new scenario buttons let you quickly surface patterns like shared pipeline or job names across projects. {width="270" height="56"} ### Getting started with the Hierarchy Map The Hierarchy Map shows the full organisational structure: Acme Corp → Subgroups → Projects, along with their pipelines and jobs. This is a good place to start click on project chips to select them, then switch to the Prototype to see the dashboard populated with that data. {width="281" height="172"} * Multi-select: click multiple project chips to build a custom selection. Selected chips appear in orange. * Scenarios: at the bottom of the page, try "Pipelines with the same name" or "Jobs with the same name". These filter to projects that share a common pipeline or job. When you switch to the Prototype, the drill-down table will automatically open on the relevant tab (Pipelines or Jobs), sorted by name — so you can immediately see the same name appearing across different projects. * {width="296" height="89"} * Clear button: the filter bar at the top of the map has a Clear button that resets both views back to their default state. ### Prototype interactions * **Subgroup filter:** use "View projects from" to select subgroups, ungrouped projects, or all groups. Selections here also highlight the relevant chips on the Hierarchy Map. * **Date range:** switch between Last week, Last 30 days, and Last 90 days. All metrics, charts, and table data adjust accordingly. * Default branches only: toggle this to filter to default branch pipelines only (\~26% of total runs). Metrics shift to reflect the reduced dataset. * **Pipeline Health Trend:** hover over data points to see a popover with that day's metrics. Click to pin it. The pinned popover shows current vs previous period comparison (the dashed pink line), a delta, and top contributors. Click "Filter in drill-down table" from a pinned popover to jump straight to that project. Toggle between Success rate (with previous period comparison) and Duration (Mean + P95 lines). * **Drill-down table:** switch between Projects, Pipelines, and Jobs tabs. Click any column header to sort (click again to reverse). The search bar and filter chips reflect your current selections — whether from the subgroup dropdown, hierarchy map picks, or scenario buttons. * **GitLab Duo Chat:** click the chat icon on the right rail. Select one of three suggested prompts to see how Duo could surface insights from the dashboard data contextually. * **Reset Prototype:** resets all filters, selections, and views back to the default state. ## Useful Links <table> <tr> <th> #### **Exploration inspiration** * [**Circle CI Insights**](https://youtu.be/E7qJInKSqqc) * [**DataDog CI Visibility**](https://youtu.be/_4Zg4Q2VCeI) * [**eCharts for our chats**](https://echarts.apache.org/examples/en/editor.html?c=multiple-y-axis) * [**Slack chat with Libor**](https://gitlab.slack.com/archives/CL9STLJ06/p1768913146365479) #### </th> <th> #### **Research Work** * **Problem Validation** * [**Issue**](https://gitlab.com/gitlab-org/ux-research/-/work_items/3328#note_2987622893) * [**Research summary**](https://docs.google.com/presentation/d/1fKxgZkxuFtam3lOhwiFhmQzum1s6g8-8AqmgFKAjq2w/edit?slide=id.g188dc7ea4ca_0_0#slide=id.g188dc7ea4ca_0_0) * [**Dovetail Customer Interviews**](https://gitlab.dovetail.com/projects/1Mzvx28zaRTLqTJ4ENFJ7Z/v/406zFLjIKreaT8KcKWwhue) * [**Survey Qualtrics**](https://gitlab.eu.qualtrics.com/reports/RC/view/SV_8jGSnMmaJM87bsq/696558d62039790008541457) * [**Use Veethika example for merge trains once ready**](https://gitlab.com/gitlab-org/gitlab/-/work_items/464774) * **Solution Validation** * [**Issue**](https://gitlab.com/gitlab-org/ux-research/-/work_items/3759) * [**Research summary**](https://docs.google.com/presentation/d/1N7GCdyusFmbxmirOqsIIAibfPrpU1zanCtId4-4KUFs/edit?slide=id.g3aaebe04ec3_0_192#slide=id.g3aaebe04ec3_0_192) * [**Dovetail Customer Interviews**](https://gitlab.dovetail.com/projects/4SpnSbAnRV42zSnkq9bK27/readme) </th> </tr> </table> ####