No 2026 DORA Annual Report: Google's DORA Pauses Its Annual Survey

DORA, the Google Cloud research programme behind the annual State of DevOps reports, announced on LinkedIn on 3 September 2026 that it would not run its annual survey this year and that there would be no 2026 State of AI-assisted Software Development report, as reported by Waydev the next day and by LeadDev on 26 October. The 2025 edition, built on nearly 5,000 respondents, found 90 percent of developers using AI at work. It may now be the last large independent, survey-based benchmark of how AI affects software delivery. For a field that has spent years arguing with DORA's charts, losing the annual dataset in the year AI changed how code gets written is bad timing.

What happened

According to Waydev's summary of the announcement, DORA framed the change as the next step in its evolution rather than an ending, and as part of a move towards qualitative and observational research. The annual survey, the annual public dataset and the yearly report are going. DORA as an organisation, its community of more than 5,000 members, its core model and its AI Capabilities Model continue.

LeadDev's reporting, by Jennifer Riggins, adds detail. The team behind the research sits within Google Cloud. Nathen Harvey, described as until recently the DORA lead, told LeadDev that "the 2025 report was our most successful report that we've had to date". He is joining Laura Tacho's team at AWS, where the pair will lead what LeadDev calls builder research; Tacho described it as "building a research group that respects what came before". A Google spokesperson said the company recognises "the importance of the research that DORA has provided to the developer community" and is "currently exploring how we can best support the future needs of developers with meaningful insights." Google did not give a reason for stopping the survey.

DORA's survey questions remain freely available on dora.dev, so teams can still run them internally.

The numbers

What DORA published in its last three cycles, which is now the reference set for anyone citing DORA on AI:

I covered these findings in more depth in DevOps Still Matters in 2026, which noted in early October that dora.dev listed no 2026 annual report.

Why it matters

DORA's value was never the four metrics themselves. You can compute deployment frequency from your CI logs. Its value was an independent, large, annual sample that let a team compare itself with an industry distribution, and let the industry test claims about practices against outcomes year on year. The 2024 finding that AI adoption correlated with lower delivery stability, and the 2025 framing of AI as an amplifier of whatever an organisation already does well or badly, were influential because they came from a neutral, repeated survey rather than a vendor.

The timing also invites questions. LeadDev quotes Sam Newman suggesting that the AI findings "don't actually help Google's products." That is one technologist's reading, not an established fact, and Google has not said anything of the kind. But the 2025 research did produce mixed results on AI's return, and those results were widely cited by people sceptical of AI coding claims.

What sceptics say

The case that this is less of a loss than it sounds is reasonable. DORA's survey was always self-reported, with the usual problems: self-selection, people rating their own teams, and changes in methodology between years that made trend lines harder to read than the charts suggested. The 2026 ROI report was a model, not a survey, which suggests DORA was already moving away from the annual-survey format. Qualitative and observational research may be better suited to a period when the tools change every quarter.

There are also other sources. Engineering analytics vendors publish telemetry-based studies (Waydev, which reported this announcement, is one), and developer surveys from Stack Overflow and JetBrains continue. Their limitation is the one DORA did not have: most come from companies that sell the measurement or the tools being measured.

What it means for engineering and DevOps teams

My take, as opinion: losing the external benchmark matters less than losing the habit it encouraged. Many teams cited DORA without measuring anything themselves. If the annual report stops, the useful response is to measure your own delivery, not to look for a replacement chart to quote.

Practical steps:

• Instrument the four keys from your own systems. Deployment frequency, lead time, change failure rate and time to restore can come from Git, CI and your incident tool. I wrote a step-by-step guide: How to Measure DORA Metrics From GitHub.
• Run the DORA survey questions internally. They are free on dora.dev. A short survey twice a year gives you a trend for your own organisation, which is more useful than an industry median anyway.
• Separate AI-authored changes. Tag pull requests from agents and compare their change failure rate and review time with human ones. That is the question the 2026 report would have tried to answer, and you can answer it for your codebase.
• Be careful with vendor benchmarks. Treat telemetry studies from tool vendors as useful but interested. Check the sample, who was excluded, and whether the metric is output (lines, PRs) or outcome (stable releases).

The metric definitions are explained in DORA Metrics Explained (2026).

FAQ

Is there a 2026 DORA report?

Not an annual one. DORA announced on 3 September 2026 that it would not run its annual survey or publish a 2026 State of AI-assisted Software Development report. It published a separate, model-based ROI of AI-assisted Software Development report in April 2026.

Is DORA shutting down?

Not according to the announcement as reported by Waydev and LeadDev. DORA says it will continue with qualitative and observational research, and its community, core model and AI Capabilities Model remain. The annual survey, public dataset and yearly report are what stop.

Can I still use DORA metrics?

Yes. The four keys are definitions, not a product, and can be measured from your own delivery systems. DORA's survey questions are still published for teams to run internally.

Related: DevOps Still Matters in 2026, DORA Metrics Explained (2026) and How to Measure DORA Metrics From GitHub.