How To Increase Efficiency With DORA

Focusing on DORA metrics increases developer performance. This is what the Google State of Devops report tells us, but does this claim stack up in the real world? We’ve taken a look at how Zoopla applies the DORA framework, and the results are compelling.

What is DORA?

DORA (DevOps Research and Assessment) is a research program famed for their ‘State of DevOps’ annual report which covers over 7 years of research and data from over 32,000 professionals. They have created a framework for maximizing developer performance, which is the bedrock of good DevOps practice in 2022.

Want a quick summary of their latest (45 page!) report? We do the hard work for you in our recent blog post.

So, this is all great in theory, but how can DORA be applied in the real world?

Metrics that matter

Zoopla has taken the 45 pages of information and data and boiled it down to its core, the metrics that matter. They started their transformation through understanding, baselining and tracking four DORA ‘delivery’ metrics:

1. Deployment frequency

Zoopla was looking to limit the amount of code going to production and therefore increase the frequency of deployments so the value was realized sooner by their users.

Baseline: 1 deployment a week

2. Lead time for changes

Here they were looking to optimize work in progress by reducing the number of blockers faced by their developers.

Baseline: 10 days

3. Time to restore services

Here they were looking to improve the speed of incident recovery by investing in SRE best practices and automation, with the aim of shortening the time required to remove any customer impact.

Baseline: not measured historically

4. Change failure rate

Finally, Zoopla was looking to reduce the percentage of changes to production that resulted in problems for their services, and subsequently required a rollback, hotfix, fix forward or patch. The aim here was to increase the quality of their releases.

How to collect DORA metrics

Zoopla had to follow a quite manual and disparate process to collate the data needed for the 4 metrics, sometimes using multiple systems for a single metric. They collected this information from Gitlab, Jenkins and Blameless using APIs and a standardized data schema they had to create. They then surfaced the data in a Google spreadsheet.

Sounds cumbersome? We think so!

Is tracking DORA metrics useful?

Yes, and the uplift in performance Zoopla experienced speaks for itself. So, we will let the data do the talking here:

What does this mean?

They are shipping value to their users over 285 times more often.

Their lead time for changes is over 5 times quicker, meaning their developers are facing far fewer blockers.

They are now focussing on how quickly they restore services in an empirical way.

Their code quality is significantly better (over 12 times, based on change failure rates)

Part of the key to Zoopla’s success with this initiative was ensuring this data was front and centre. Their dashboard is a central information radiator for their senior engineering management team but is also included in monthly town hall meetings and operations meetings. It is also used actively on the ground by individual developers and teams who are encouraged to retro their metrics to find ways to leverage them and improve.

Can we go beyond DORA’s delivery metrics?

DORA as a framework is actually a whole lot more than just productivity metrics. It also notably focuses on reducing burnout amongst developers. With a recent study showing that 83% of developers feel a degree of burnout, it is a significant business and societal issue. We also know that productivity and wellbeing are intrinsically linked, so the importance of reducing burnout both personally and in our teams is integral.