The four DORA metrics - deploy frequency, lead time for changes, change failure rate, and time to restore - came out of years of research long before coding agents existed. It would be reasonable to expect them to feel dated in the AI era. They don't. If anything, they've become more useful, because they resist exactly the kind of gaming that AI makes easy.
Why they still work
Their durability comes from a single design choice: they pair speed with stability. Two of the metrics measure how fast you move (deploy frequency, lead time) and two measure whether you can trust the movement (change failure rate, time to restore). You can't juice one pair without eventually paying for it in the other. That built-in tension is what keeps a team honest.
| Metric | Measures | Elite-ish target |
|---|---|---|
| Deploy frequency | Speed | On-demand / multiple per day |
| Lead time for changes | Speed | Under a day |
| Change failure rate | Stability | Under ~15% |
| Time to restore | Stability | Under an hour |
Why they resist AI gaming
Vanity metrics - lines of code, PR counts, commits - collapse the moment an agent can produce them on demand. DORA doesn't, because it measures the system's behavior in production, not how much text got generated. An agent can write a thousand lines in a minute; it cannot lower your change failure rate by doing so. That makes DORA a stable floor to build on while everything above it is in flux.
Where they fall short
DORA tells you about the system, not the change. It won't tell you whether a specific pull request was high quality, whether a particular service is dragging the average, or - critically - whether AI-authored work is more failure-prone than human work. It's a team-level thermometer, not a diagnosis.
That's the gap to fill deliberately. Keep DORA as your foundation and layer two things on top:
- Per-change quality scoring, so you can see which PRs and which authors move the failure rate.
- AI-vs-human attribution, so "did the agents help" stops being a vibe and becomes a chart.
Use them as a floor, not a ceiling
Treat DORA as the non-negotiable baseline every team reports, then let richer per-change measurement explain the why behind the trend. Want the numbers for your team first? The DORA Metrics Calculator gives you a quick classification, and the full guide walks through each metric and how to start collecting it.
Measure what you ship - for real.
Put your team’s AI web-development on the outcome loop.
Get early access