Our delivery.
We help clients measure their software delivery. Here are our own numbers for this site — the four DORA metrics, computed from the repository and the hosting every time the site is built. Nothing hand-written.
Last 88 days, from 20 July 2026. Computed on 16 October 2026, when the site was built.
Deployments per week
Code changes are pushes to main. Rebuilds are when Sanity publishes content and triggers a new build of the same code.
Hover a week for its numbers.
Latest deployments
| When | What | Kind | Lead time |
|---|---|---|---|
| 15 Oct | 35a5d30 feat(konsulttjanster): three defined entry offers with what is included and how they are priced | Rebuild | – |
| 7 Oct | 35a5d30 feat(konsulttjanster): three defined entry offers with what is included and how they are priced | Rebuild | – |
| 5 Oct | 35a5d30 feat(konsulttjanster): three defined entry offers with what is included and how they are priced | Code | 2 min |
| 5 Oct | 918de89 docs(changelog): the DeLaval vCPU label names Fargate | Rebuild | – |
| 5 Oct | 918de89 docs(changelog): the DeLaval vCPU label names Fargate | Code | 2 min |
| 5 Oct | b3b2fa3 feat(konsulttjanster): "What do you need help with?" selector above the services (#30) | Rebuild | – |
| 5 Oct | b3b2fa3 feat(konsulttjanster): "What do you need help with?" selector above the services (#30) | Code | 4 min |
| 5 Oct | f57c2f1 feat(startsidan): results strip under the introduction, slot for a customer quote | Code | 2 min |
| 5 Oct | 31aa4ee fix(fallstudier): leave out outcome cards that have no number | Code | 2 min |
| 5 Oct | d709b2d docs(changelog): new post on the day's twelve changes to the notebook | Rebuild | – |
How we count
- Deployment
- Every time a new version of the site goes live in production on Vercel — recorded as a successful deployment status in GitHub. A push to main gives one; a publish in Sanity gives a rebuild of the same code.
- Lead time
- From when a commit was written (author date) until the first deployment containing it went live. For squash-merged pull requests, from the branch's first commit.
- Failed change
- A build that failed in production, or a deployment whose commit was later reverted on main.
- Time to restore
- From the failed deployment to the next one that went live.
The source is GitHub's deployment and commit APIs for the repository, read when the site is built; nothing is fetched in your browser. The window is thirteen Swedish calendar weeks. The site has no monitoring that could catch a failure in production after the fact, so a failure nobody reverted is not counted — that is the honest limitation.
Why we publish this
We help clients measure and improve exactly these four numbers — the DORA metrics — so it would be odd not to publish our own. This site is a small, single-developer project with a static build; the figures are not a benchmark for a product team, but the mechanism is the same one we set up for clients: measure from what actually happened, publish the numbers, let the trend drive the work.
Want the same for your own delivery? That is what Delivery Engineering is about. The wait-time calculator puts a price on the minutes between commit and verdict.