Cohort analysis

The Cohort analysis report compares engineering metrics side-by-side across user-defined segments of your organization. Define up to three cohorts based on team attributes, then see how each group performs relative to a baseline.

This report helps teams:

  • Compare performance across segments — Measure how different teams, roles, locations, or any other attribute-based group stack up on key engineering metrics.
  • Evaluate the impact of changes — See whether a new process, tool, or organizational change is moving the needle for a specific group compared to the rest of the organization.
  • Identify areas for investment — Spot which segments are underperforming on cycle time, throughput, or developer experience so you can target support where it’s needed most.
  • Benchmark against industry — Overlay industry P50/P75/P90 benchmarks on any metric to see how each cohort compares externally.

Cohort analysis report in DX

Setting up cohorts

The report starts with two groups: a Baseline and Cohort 1. You can add up to two more (three total in addition to the baseline) using the Add cohort button.

  • Baseline — The reference group that all other cohorts are compared against. By default this is “All attributes across all users,” but you can customize it to any attribute set.
  • Cohort 1–3 — Comparison groups. Select attributes to define each cohort’s membership.

Defining a cohort with attributes

Each cohort is defined by selecting one or more attributes from your organization’s attribute groups (e.g., Team, Location, Role). Attributes support both inclusion and exclusion:

  • is / is any of — Include users matching the selected attribute values.
  • is not / is not any of — Exclude users matching the selected attribute values.

You can combine multiple attribute groups within a single cohort for more targeted comparisons (e.g., “Team is Engineering and Location is not Remote”).

Metrics

Each metric is displayed as a tile in a responsive grid. Use the tiles dropdown (grid icon) to choose which metrics to display. Your tile selection is saved across sessions.

Available metrics

Metric Source Description
TrueThroughput™ / PR throughput System Complexity-weighted throughput (when AI tools are enabled) or raw PR count per engineer per week
PR revert rate System Percentage of PRs that are reverted
PR cycle time System Time from first commit to PR merge
PR size System Lines of code per PR
Review comment density System Comments per PR
Review pushback System Reviews per PR before merge
Review turnaround System Time from review request to first review
Story points completed System Points completed per contributor (requires Jira connection)
Innovation ratio System Percentage of work allocated to feature development (when AI tools are enabled)
Defect ratio System Percentage of work allocated to bug fixes (when AI tools are enabled)
PR throughput (self-reported) Snapshot Self-reported throughput from developer surveys
DXI Snapshot Developer Experience Index score
Change failure (self-reported) Snapshot Self-reported change failure percentage
Allocation (self-reported) Snapshot Self-reported innovation ratio
Conditional drivers Snapshot Additional self-reported metrics based on your snapshot configuration (e.g., velocity, code maintainability, change confidence, focus time)