DORA Metrics

After years of research, Google’s DevOps Research and Assessment (DORA) team identified four key metrics for evaluating a team’s performance:

Please follow the links above to find out how to create DORA metrics in PERF.

After that please come back - below you'll find an example of converting values on DORA metrics into scores (based on benchmarks) & creating a composite score metric, which shows averages of four key metrics scores. This will give you a general idea of the status & trend of DORA metrics for your teams.

The pre-requisite for it is to have all 4 aggregated data sets for all 4 DORA metrics already available.

We will convert the value of each DORA metric for each team into scores based on RAG thresholds and then take the average: Composite score = (Deployment Frequency score + Lead Time for Changes score + Mean Time to Recovery + Change Failure Rate)/4.

RAG → Score conversion:

RED (Low) = 0
AMBER (Medium) = 1
GREEN (High) = 2

RAGs are taken from 2022 Google's State of DevOps Report: https://services.google.com/fh/files/misc/2022_state_of_devops_report.pdf

Thresholds on a composite metric could be not so strict - for example:

Here is the aggregated metric's code:

with dora_scores as --Calculating DORA scores
(
select a.timeline, a.unit_name,
case when deployment_frequency<1 then 0
  	 when deployment_frequency between 1 and 4 then 1
  	 else 2 end
  as dp_score,
case when lead_time_for_changes<7 then 2
  	 when lead_time_for_changes between 7 and 30 then 1
  	 when lead_time_for_changes is null then null
  	 else 0 end 
  as ltfc_score,
case when mean_time_to_recovery<1 then 2
  	 when mean_time_to_recovery between 1 and 7 then 1
     when mean_time_to_recovery is null	then null
     else 0 end 
  as mttr_score,
case when change_failure_rate<15 then 2
  	 when change_failure_rate between 15 and 30 then 1
  	 when change_failure_rate is null then null
     else 0 end 
  as cfr_score
from DORA_Deployment_Frequency a
join DORA_Lead_Time_for_Changes b on a.timeline=b.timeline and a.unit_name=b.unit_name
join DORA_Mean_Time_to_Recovery c on a.timeline=c.timeline and a.unit_name=c.unit_name
join DORA_Change_Failure_Rate d on a.timeline=d.timeline and a.unit_name=d.unit_name
)

select 
timeline, round(avg(score),2) as "Composite DORA Score"
from
(
select timeline, avg(dp_score) as score from dora_scores group by timeline
union all
select timeline, avg(ltfc_score) as score from dora_scores group by timeline
union all
select timeline, avg(mttr_score) as score from dora_scores group by timeline
union all
select timeline, avg(cfr_score) as score from dora_scores group by timeline
) as temp
group by timeline
order by timeline asc

Here is the drill-down code with High/Medium/Low levels for each metric & team:

select unit_name as "Team", 
case when dp_score<1 then 'Low' 
	 when dp_score>=1 and dp_score<2 then 'Medium'
     when dp_score>=2 then 'High' else null end
as "Deployment Frequency",
case when ltfc_score<1 then 'Low' 
	 when ltfc_score>=1 and ltfc_score<2 then 'Medium'
     when ltfc_score>=2 then 'High' else null end
as "Lead Time foe Changes",
case when mttr_score<1 then 'Low' 
	 when mttr_score>=1 and mttr_score<2 then 'Medium'
     when mttr_score>=2 then 'High' else null end
as "Mean Time to Recover",
case when cfr_score<1 then 'Low' 
	 when cfr_score>=1 and cfr_score<2 then 'Medium'
     when cfr_score>=2 then 'High' else null end
as "Change Failure Rate"
from dora_scores
where timeline=clicked_x_value
order by unit_name asc

Aggregated custom metrics can be added on dashboards for upper units in hierarchy as well as on the team's dashboard itself - in this case metric & the pop-up that appears on the click will contain this particular unit's data only: