Skip to content

Insights

Insights is a view in the app. It scores two DORA metrics from the Evidence Events already stored for your organisation. Nothing is written to a separate metrics store: opening the page, or calling the API, computes the numbers for the window you pick.

The default window is the last 30 days. Anyone who can read events can open Insights.

A production deploy is deploy.succeeded whose environment is prod or production (the environment field, the environment identity, or an env: tag). Other environments are not scored. k8s.rollout.succeeded counts, because it is the same deploy event.

Deployment frequency

How often successful production deploys happen in the window.

Each deployment counts once. When the event has a deployment identity, later events for that same deployment and service are ignored. The rate is the count divided by the length of the window.

The page shows that rate, a bar for each day (or each week, when the window is longer than 62 days), and the count by service. Chart days follow the timezone of the browser.

BandSuccessful production deploys
EliteMore than once per day
HighBetween once per day and once per week
MediumBetween once per week and once per month
LowFewer than once per month

A window with no production deploys has no band.

Lead time for changes

How long a change takes to reach production.

For each successful production deploy in the window, lead time starts at build.commit.pushed for that commit. If that event is missing, it starts at the earliest other evidence of the same commit that is not itself a deploy. The commit and the deploy must name the same service when both events have one.

Deploys with no earlier evidence are left out of the median. They still count toward deployment frequency. Commit evidence is included from up to 90 days before the window, so a deploy near the start can still see the commit that shipped.

The number on the page is the median of those lead times. An even number of samples averages the two middle values.

BandMedian lead time
EliteLess than one hour
HighLess than one day
MediumBetween one day and one week
LowMore than one week

These bands are the ones from Accelerate.

API

The app reads GET /insights/dora. Omit from and to for the last 30 days. Set both, as RFC3339 timestamps, to choose a window. tzOffset is minutes east of UTC and only aligns the chart buckets.

http
GET https://api.evitrus.io/insights/dora?from=2026-08-25T00:00:00Z&to=2026-09-24T00:00:00Z&tzOffset=120
Authorization: Bearer …

The response includes deployCount, perDay, frequencyLevel, leadTimeMs, leadTimeLevel, leadSamples, leadUncovered, byService, and series. Levels are elite, high, medium, or low, or null when there is nothing to score.

The caller needs the events.read permission.

Evitrus — observability for auditability