Deployment Frequency #
Part of the Dora Metrics documentation. For the metric cards, the graph types, the Application Metric Table and service-level metrics, see the Dora Metrics Guide. For how to open the dashboard and choose a date range, see the Maturity Insights Overview.
What is Deployment Frequency? #
Deployment Frequency measures how often your teams successfully release changes to production. It is one of the key DORA metrics and reflects the speed of your delivery process: teams that deploy frequently ship improvements and fixes to users faster, in smaller and lower-risk batches.
In BuildPiper, Deployment Frequency counts the successful production deployments, including hotfixes, completed during the selected period. Rollbacks and failed deployments are not counted. A higher value is better.
To view it, select the Deployment Frequency card on the Dora Metrics dashboard. The graphs below the cards then show its trend, thresholds, maturity levels, top and weakest applications, and release activity. This guide explains how the metric is calculated and how to read each graph.
How Deployment Frequency is calculated #
You can also see this explanation on the dashboard: click How it Works in the Thresholds Legend.
What is counted #
| Counted | Not counted |
|---|---|
| Successful production deployments | Rollback deployments |
| Hotfix deployments | Failed deployments |
Deployment Frequency uses the Per Period aggregation (the deployments in the selected date range are added up) and the Highest rating rule.
How the level is calculated #
- 1
BuildPiper counts the eligible deployments in the selected date range.
- 2
Each level’s threshold is defined for a standard evaluation period, configured in Maturity Settings (for example, 7 days). It is scaled to the number of days you selected and to the number of services that have data.
- 3
The deployment count is compared with the scaled thresholds, and the highest level reached is shown: Elite, High, Medium or Low.
The scaled threshold is calculated as:
Scaled threshold = (Base deployment count ÷ Evaluation period) × Selected period × Contributing services
| Term | Meaning |
|---|---|
| Base deployment count | The number of deployments a level needs in one evaluation period. For example, Elite needs more than 10 deployments per week. |
| Evaluation period | The standard period the thresholds are defined for, in days. For example, 7. |
| Selected period | The number of days in the date range you selected. |
| Contributing services | The number of services that have data. This is 1 for a single service. |
Worked example #
Elite needs more than 10 deployments per week (evaluation period of 7 days), and you select a 90-day period.
| Level | Calculation | Needed for Elite |
|---|---|---|
| Service | (10 ÷ 7) × 90 × 1 | More than 128 deployments |
| Application with 2 services that have data | (10 ÷ 7) × 90 × 2 | More than 257 deployments |
| Organization with 5 services that have data | (10 ÷ 7) × 90 × 5 | More than 642 deployments |
Good to know: Because thresholds grow with the number of services that have data, an application with many services needs many more deployments to reach the same level. That is why an application can show a high deployment count and still be rated Low.
Deployment Frequency graphs #
Each graph type is introduced in “Metric graphs” in the Dora Metrics Guide. This section shows what each graph looks like for Deployment Frequency and how to read it.
Deployment Frequency Trend #
What it shows: When releases occurred during the selected period.
Example: Deployment frequency was about 42 in the week of 25 September–1 October, dropped to about 6 the next week, rose again to about 16 in 9–15 October, and then fell to almost zero by 23–24 October.
Use it to: Identify busy release periods, quieter periods and overall delivery patterns.
Thresholds Legend #
What it shows: The number of deployments needed to reach each level for the selected period. These thresholds are scaled to the selected period and the number of services that have data (see “How Deployment Frequency is calculated”).
Example: Elite needs more than 360, High needs 270–360, Medium needs 180–270, and Low is 180 or less. Deployment Frequency is measured as the number of successful deployments per day, and higher is better.
Use it to: Know exactly how many deployments are needed to reach the next level.
Metal Rating Distribution #
What it shows: How applications are spread across the Deployment Frequency levels.
Example: 6 of 7 applications are Low (85.71%) and 1 is Medium. None are Elite or High.
Use it to: Understand how release frequency is spread across your applications.
Top Performing Applications #
What it shows: The applications with the strongest Deployment Frequency, with their values.
Example: The chart lists ot-demo-ms (0.8), coe-softwares (2.1), bp-prod-play (4.3), PMS (6.7) and ToolHub (9.4).
Use it to: Identify top performers and learn from their practices.
Need Attention Applications #
What it shows: The applications whose Deployment Frequency needs improvement, with their deployment counts and levels.
Example: coe-solutions (353), SpendSmart (26), coe-softwares (19), rems (11) and PMS (7) are all rated Low. coe-solutions has the most deployments but is still Low, because its threshold is scaled by its large number of services (see “How Deployment Frequency is calculated”).
Use it to: Build a short list of applications to improve first.
Application Release Activity #
What it shows: Release activity for each application over time. Each row is an application and each column is a time interval. Each cell shows the number of releases, and darker cells mean more releases.
Scroll sideways to see more intervals, and use the page numbers at the bottom to see more applications.
How to read it: A more even pattern suggests regular releases, while gaps may indicate periods with little or no release activity.
Example: coe-solutions released 155 times in 21 June–20 July, then 98, 72 and 28 in the following months, and none in 19–24 October. coe-softwares released 17 times in 21 June–20 July and only 2 times after that.
Use it to: Spot applications that release irregularly or have stopped releasing.