Developer productivity metrics
for AI-assisted teams.

What are developer productivity metrics?

Developer productivity metrics describe how well an engineering team turns work into software that ships. No one number captures it: the authors of the SPACE framework write that productivity “cannot be measured by a single metric or dimension.” The frameworks most teams start from:

• DORA: software delivery performance (change lead time, deployment frequency, failed deployment recovery time, change fail rate and deployment rework rate).
• SPACE: five dimensions from GitHub and Microsoft Research (ACM Queue, 2021). They are satisfaction and well-being, performance, activity, communication and collaboration, and efficiency and flow.
• DX Core 4: DX's framework that encapsulates DORA, SPACE and DevEx in four dimensions (speed, effectiveness, quality and business impact).

How AI coding changes what to measure

Coding agents make a change cheaper to write, so activity counts rise whether or not more ships. And how fast AI feels is a poor guide. In METR's randomized study of 16 experienced open-source developers working on 246 issues (July 2025), developers took 19% longer when allowed to use AI tools, and afterwards still believed AI had sped them up by 20%.

In February 2026 METR said its follow-up could no longer measure the effect reliably: 30% to 50% of developers said they held back tasks they didn't want to do without AI. Its raw estimate for newly recruited developers was 4% less time per task, anywhere from 15% less to 9% more.

DX's AI Measurement Framework measures AI code assistants and agents on three dimensions: utilization, impact and cost. DX measures them with system data and surveys together; Zest supplies the system side for AI coding sessions.

How to measure AI coding impact per engineer

Zest is AI coding analytics: its plugins and extension report each session from the coding agents your engineers run, and the Zest GitHub App joins those sessions to what merged.

• Utilization: which coding agents, models, skills and MCP tools each engineer uses, across Claude Code, Cursor, Codex, GitHub Copilot Chat, Grok Build and Hermes.
• Cost: input and output tokens per session, with an estimated cost from model list prices.
• Impact: the pull request each session led to, and PR throughput (merged pull requests per week, averaged over four weeks).
• Speed: PR cycle time (median and p75 hours from opened to merged, week over week), through Ask Zest or Zest's MCP server.

From AI coding metrics to a better next week

Each evening Zest writes every engineer's standup from that day's sessions and pull requests, so nobody types a status update.

Once a week (Fridays by default) the Team Standup lists the AI stack worth sharing and up to three moves to ship more, so the practices behind one engineer's merged work can spread to the rest of the team.

What Zest does not measure

Zest doesn't calculate DORA metrics or run developer-experience surveys. It doesn't ingest deployments or incidents, and it doesn't capture Copilot inline completions. Keep your CI/CD, incident and survey tools for delivery and sentiment, and use Zest for the AI coding work that feeds them.