Coding throughput as a measure of productivity

In a previous Radar, we warned against using coding throughput as a measure of productivity. As the industry pushes for AI adoption and seeks tangible metrics, we increasingly see activity-based metrics such as the percentage of code generated by AI or tokens consumed treated as badges of honor. Although such metrics may indicate adoption levels, they’re poor proxies for real productivity gains.

To measure meaningful productivity, we continue to recommend established frameworks such as DORA metrics and collaboration quality, along with tying improvements directly to business outcomes. Treating activity metrics as targets rather than signals risks falling prey to Goodhart's law. We’ve seen developers generating AI busywork simply to hit these targets.

AI coding assistants are delivering real productivity gains and are rapidly becoming standard developer tooling. However, we’re increasingly seeing organizations measure success using superficial indicators such as lines of code generated or the number of pull requests (PRs). When these coding throughput metrics are used in isolation, they can negatively shape employee behavior. The result is often a flood of poorly aligned code that slows reviews, harms delivery throughput and introduces security risks. Cycle times increase as engineers raise PRs filled with insufficiently reviewed AI output, leading to repeated back-and-forth with reviewers. These metrics fail to capture the residual effort required to adapt AI-generated code to a team's architecture, conventions and patterns.

More meaningful leading indicators exist, such as first-pass acceptance rate — how often AI output can be used with minimal rework. Measuring this exposes hidden effort and makes improvement actionable: teams can refine prompts, improve priming documents and strengthen design conversations to progressively increase acceptance over time. This creates a virtuous cycle in which AI output requires less correction. First-pass acceptance also connects naturally with DORA metrics: lower acceptance rates tend to increase change failure rates, while repeated iteration cycles extend lead time for changes. As AI assistants become ubiquitous, organizations should shift focus away from coding throughput alone toward metrics that reflect real impact and delivery outcomes.