What AI engineering teams are actually measuring

Agile metrics built around velocity and story points are being gamed by AI coding tools that inflate estimates and accelerate task completion without improving shipped outcomes. The reliable signals now are cycle time, deployment frequency, code review velocity, and deploy-to-fix latency.

Cycle time tracks how quickly code moves from commit to production (typically 2 to 7 days). Lead time tracks how quickly a feature moves from request to production (typically 2 to 8 weeks). AI tools reduce the first but rarely touch the second, and the widening gap between them reveals where your real bottleneck sits.

Developer productivity metrics built around lines of code stopped working the moment AI coding assistants entered the workflow. Cycle time, code review velocity, deploy-to-fix latency, and velocity per FTE are the replacement signals: infrastructure-measured, correlated with business outcomes, and resistant to AI-driven inflation.

A pull request is a version control feature that lets developers propose code changes to a repository and request review before merging to the main branch. It is the primary quality gate in modern software development, and AI-generated code makes careful review a core engineering skill rather than a workflow step.

Four signals measure AI coding tool ROI reliably: cycle time reduction, story-point velocity per FTE, deploy-to-fix latency, and code review time compression. Combined, they isolate whether AI tooling spend is producing engineering outcomes or just inflating activity dashboards.

DORA metrics are four engineering performance measures validated across thousands of organisations as the industry standard for benchmarking software delivery performance. In 2026, they remain necessary but no longer sufficient.