Which five numbers does a DORA metrics tool report?
A DORA metrics tool reports software delivery as five numbers in two groups, throughput and instability, according to DORA's own guide (last updated January 5, 2026, and read again on November 16, 2026), up from the original four keys. Many tools still compute the original four, so check which set a vendor means.
| Metric | Group | DORA's definition | The data it needs |
|---|---|---|---|
| Change lead time | Throughput | Time for a change to go from committed to deployed in production | Commit times and production deploy times |
| Deployment frequency | Throughput | Number of deployments over a period, or time between them | Production deployment events |
| Failed deployment recovery time | Throughput | Time to recover from a deployment that fails and needs immediate intervention | Failed deployments and when they were fixed |
| Change fail rate | Instability | Share of deployments that need immediate intervention | Deployments linked to incidents or rollbacks |
| Deployment rework rate | Instability | Share of deployments that are unplanned and caused by a production incident | Which deployments were unplanned fixes |
Where does a DORA metrics tool get its data?
A DORA metrics tool gets its data from three feeds: Git for commits, your CI/CD system or a deployment API for production deployments, and an incident tracker for failures. Git alone gives you time to merge, not time to production, and that is where setups go wrong, so ask any vendor how it decides what counts as a production deployment and an incident.
The two open references, both read on November 16, 2026, show the choices. GitLab counts successful deployments to an environment in its production tier, or one named production or prod, and uses GitLab incidents; pull-based deploys through its agent create no deployment record unless you send one through the Deployments API, and duplicate incidents inflate the change failure rate. Apache DevLake reads deployments from five CI systems (Jenkins, GitHub Actions, GitLab CI, Bitbucket and Azure Pipelines), incidents from Jira or GitHub issues, and anything else through webhooks.
Which DORA metrics tools are free or open source?
Three DORA metrics tools cost nothing on top of what a team may already have, as read on November 16, 2026: Apache DevLake (open source, if you host it), GitLab's DORA metrics (built into the Ultimate tier) and Swarmia's free plan for companies with fewer than 10 developers. Datadog's version comes with Datadog, and Four Keys is archived.
Apache DevLake
Apache-2.0 licensed and still active (its GitHub repository had a push on November 16, 2026). You install it with Docker Compose or Helm, and its DORA dashboard covers four metrics, with failed deployment recovery time in place of the older mean time to restore. Free, but you run it.
GitLab DORA metrics
On GitLab.com, Self-Managed and Dedicated, but only in the Ultimate tier. It computes four metrics: deployment frequency, lead time from merge to production, time to restore service from incidents, and change failure rate. No extra tool if you already pay for Ultimate.
Swarmia's free plan
Swarmia's free plan has everything in Standard except HRIS integrations and Swarmia AI, for companies with fewer than 10 developers. Its DORA view lists change lead time, deployment frequency, change failure rate and mean time to recovery.
Datadog DORA Metrics
Datadog's DORA Metrics computes the four key metrics (deployment frequency, change lead time, change failure rate and failed deployment recovery time) from deployment events you send it. Its documentation doesn't give a price.
Four Keys
The open-source project from the DORA team that many older guides still recommend is archived on GitHub, with no code change since January 2024. Don't start a new setup on it.
Which paid platforms compute DORA metrics?
Nine of the ten paid platforms in this site's comparison of engineering intelligence platforms compute DORA metrics, from $20 a developer a month (Typo Starter, billed annually) up, on prices read November 16, 2026 for the engineering intelligence platform pricing data. They differ less on the metrics than on price, where the data comes from and what else you get with it.
| Platform | Where DORA sits | List price |
|---|---|---|
| Typo | DORA Metrics on Starter, with flow and quality metrics | $20 a developer a month, annual |
| Swarmia | Productivity & AI impact module, or Standard | $23 for the module, $45 Standard; free under 10 |
| Plandek | Among 500+ metrics, with a ready-made DORA board | $25 to $55 a contributor a month |
| LinearB | DORA metrics on both plans | $29 a user a month, 50-user minimum |
| Allstacks | DORA, SPACE and Flow metrics | $400 a contributor a year |
| DX | A DORA metrics report in its docs | Quote; AWS listing from $15,000 a year |
| Jellyfish | Developer Productivity module (DORA, SPACE, AI metrics) | Quote; AWS bundle $30,800 a year for 15 seats |
| Waydev | A DORA Metrics product; the Pro plan lists CI/CD reports | Custom quote, annual |
| Faros | Its February 2026 blog says it tracks all five, with unlimited history | Consumption-based quote |
The tenth, Zest, doesn't compute DORA metrics: it reads coding-agent sessions and links them to merged pull requests, which answers how AI tools get used rather than how often you deploy.
How do you pick a DORA metrics dashboard?
A DORA metrics dashboard is worth picking on how it defines a deployment and an incident, how many of the five metrics it shows, and its price at your head count: Swarmia is free under 10 developers, and LinearB's Essentials plan bills at least 50 users (pricing pages read November 16, 2026).
- Check what it calls a deployment. A tool that counts merges to main as deployments will flatter your lead time. Look for production environments or a deployment API.
- Find out where incidents come from. Change fail rate and recovery time are only as good as the incident feed: Jira, PagerDuty, GitHub issues or GitLab incidents.
- Ask whether it has five metrics or four. Rework rate is new, and most dashboards don't show it yet.
- Keep it at team level. DORA describes a delivery system. Ranking people by deployment counts invites gaming.
- Price it at your head count. Under 10 developers, Swarmia is free. Past 50, LinearB's minimum stops mattering. With Gerrit or a strict data policy, look at DX or Waydev; for the full buying checklist, see how to choose an engineering intelligence platform.
Leaving a tool you already have? Waydev pricing sets the options against Waydev's plan limits.
How was this page checked?
We read DORA's guide, GitLab's, Apache DevLake's and Datadog's DORA documentation, and the GitHub records for DevLake and Four Keys, on November 16, 2026. Every price on this page was read on the vendor's own pricing page or AWS Marketplace listing on November 16, 2026; the same figures, each with its URL, are on the engineering intelligence platform pricing data page and in its CSV. Where we say a product lacks something, we mean its public pages don't list it. Corrections: support@winding.ai.
Published by Winding AI, Inc., the company that makes Zest (AI coding analytics for engineering teams), one of the products compared here; Winding AI, Inc., 1887 Whitney Mesa Drive #5338, Henderson, NV 89014. We make money when teams pay for Zest; no vendor paid to be listed and no link earns a commission.