AI's Productivity Boom Leaves Engineering Metrics Behind

Event summary

  • Harness's 2026 report reveals 89% of engineering leaders report AI-driven productivity gains, but 94% admit key metrics like code quality and burnout are missing from current frameworks.
  • 81% of developers report increased time spent on code review since adopting AI tools, with 28% seeing a 30%+ rise in manual work.
  • 31% of developer time is now spent on untracked tasks like reviewing AI-generated code and fixing bugs.
  • 54% of developers fear AI productivity data will be used for individual performance evaluations.
  • Harness recommends measuring code quality, validation time, cognitive load, and burnout alongside traditional metrics.

The big picture

Harness's report highlights a critical disconnect in the software engineering industry: while AI tools are delivering productivity gains, traditional measurement frameworks fail to capture the full picture of developer work. This visibility gap could lead to misguided investment decisions and overlooked operational costs as engineering leaders rely on outdated dashboards for multi-year AI strategies. The findings underscore the need for more comprehensive metrics that account for code quality, cognitive load, and burnout alongside traditional velocity and cycle time measures.

What we're watching

Measurement Evolution
How engineering organizations will adapt their metrics to capture both AI benefits and hidden costs.
Developer Trust
Whether organizations can bridge the perception gap between leadership and practitioners on AI productivity data usage.
AI Integration
The pace at which AI performance will be treated as a distinct discipline within engineering organizations.