Woman Configure

Where DevOps Failed and What You Should Try Instead

July 24, 2024
Roger Turnau

DevOps has the distinction of being one of the few innovations in technology that has rock-solid data to back it up. For years, the gold standard for IT performance has been encoded in the DevOps Research and Assessment (DORA) metrics: deployment frequency, lead time for changes, mean time to restore service, and change failure rate. Get these right, advocates said, and you, too, will outperform your peers. 

The State of DevOps Report: A Major Dip in Performance

The annual State of DevOps Report has long classified organizations into high, medium, and low performers so we can all keep score. Then, in 2022, the report noticed something odd. The number of low performers more than doubled in a year, from 7% to 19%. Even worse, the number of high performers dropped significantly, from 66% to 11%. 

While you can blame some of the dip on the COVID pandemic, it seems unlikely that would explain all of it. Another data point in 2022 shows that the performance dip may have been triggered by developer burnout. 

What’s Causing Developer Burnout

Haystack Analytics published a report on how developers were affected by the pandemic, and a shocking 83% said that they suffered from some level of burnout. 55% said they felt moderate to high levels of burnout, while 28% said it was just a small extent. We can argue semantics, but a small amount of burnout is still probably too much. 

The top contributor to this, according to the Haystack report, was increased workload. It appears that a lot of companies had spent a lot of time trying to increase their DORA metrics, placing the entire cognitive load of this effort on their development teams. 

After all, a big selling point for DevOps was that you could speed up time-to-market and deliver high-quality software without any trade-offs. The “no trade-offs” part of that equation was always too good to be true. But tech executives barreled ahead, overleveraging on promises of performance while simultaneously ignoring everything the DevOps community had to say about organizational culture and psychological safety. 

In short, the main failure of DevOps was not realizing that developers’ stress response is not infinitely elastic. There is only so much you can dump on the laps of developers before things start falling off and the devs start jumping ship.

A Shift in Developer Performance Metrics

It was around this time that we started to see the term “cognitive load” crop up in books and research papers. The idea was that software development was inherently complex, so we needed some way to account for the amount of mental processing required for a developer to do a new task. A routine task that has been done multiple times a day comes with a low cognitive load. An unfamiliar, non-routine task with significant risks imposes a greater cognitive load.

A new field of platform engineering emerged to help address the main issues of overwhelmed dev teams. Soon after, an even deeper understanding of developer performance started to emerge.

Shortly after formalizing her DORA metrics in 2020, well-known researcher and DevOps expert Nicole Forsgren turned around and published a less prominent set of metrics known as SPACE: 

  • Satisfaction and Well-Being
  • Performance
  • Activity
  • Communication and Collaboration
  • Efficiency and Flow

Unlike DORA, there are no formal definitions for these metrics. Rather, they show the full set of domains that need to be accounted for in order to get a holistic picture of developer performance. 

Exit DevOps, Enter DevEx

The point of DORA metrics was always to have a baseline for the technical performance of the overall organization. Since developers are the primary front-line workers whose contributions significantly impact business performance, many organizations have started to prioritize Developer Experience (DevEx)

DevEx is complex and nuanced, but is particularly concerned with improvements along the following three dimensions:

  • Flow State
  • Feedback Loops
  • Cognitive Load

The first, Flow State, is heavily influenced by the work of Mihaly Csikszentmihalyi and his seminal book Flow: The Psychology of Optimal Experience. Feedback loops are somewhat self-explanatory. If, for instance, it takes your senior developer two weeks to review your pull request, then your feedback loops need some attention. 

The core idea of DevEx is to ensure your developers have what they need to succeed, while also considering how the developers feel about, think about, and value their work. Given the balanced way in which SPACE metrics are defined, it should be much harder for managers to game them. This shift is also a great time to reassess the entire point of metrics.

The core idea of DevEx is to ensure that developers have what they need to succeed, while also considering how the developers feel about, think about, and value their work.

The moment you use metrics to reward or punish any team or employee, that is the last time you will have an honest conversation or a trustworthy metric. Any assessment of developer performance should be blameless and aimed at helping developers excel at their jobs.

Start Your DevEx Journey 

To keep up with ever-evolving customer demands and financial goals, enterprise organizations must build and release quality, reliable software. This is why creating a great developer experience has become a core focus for companies.

If you would like to start your DevEx journey, Praecipio has experts who can help you assess your developers’ performance, job satisfaction, and engagement. 

Contact our team about creating a DevEx strategy that benefits the entire organization. 

Elevate The Developer Experience

Improve Your DevEx and ROI