DORA、組織目標に合った測定フレームワークの選び方を解説 Choosing measurement frameworks to fit your organizational goals

Measuring software delivery performance has become a crowded discipline, with multiple frameworks competing for attention. DORA's latest insight piece argues that the right choice depends less on which framework is fashionable and more on what an organization is actually trying to improve.

The article walks through the strengths and intended uses of several well-known frameworks. The DORA metrics — deployment frequency, lead time for changes, change failure rate, failed deployment recovery time, and reliability — remain a widely adopted way to measure software delivery throughput and stability. The SPACE framework (Satisfaction, Performance, Activity, Communication, Efficiency) takes a deliberately broader view, recognizing that developer productivity cannot be captured by output metrics alone. The DevEx framework, more recent in origin, focuses on feedback loops, cognitive load, and flow state as proxies for the lived experience of engineers.

Rather than treating these as competing options, DORA frames them as complementary lenses. An organization wrestling with slow or unstable delivery may benefit most from the DORA metrics, while one facing burnout, attrition, or unclear engineering priorities may find SPACE or DevEx more illuminating. The piece cautions against picking a single metric and optimizing for it, since narrow targets tend to invite gaming and obscure the underlying system dynamics that frameworks are meant to reveal.

The broader context here matters. Interest in developer productivity measurement has surged in the past few years, fueled both by economic pressure on engineering budgets and by the rise of AI coding assistants whose impact organizations are eager to quantify. Vendors such as GitHub, Atlassian, LinearB, Jellyfish, and Swarmia have built products around these frameworks, and a McKinsey article proposing its own developer productivity metrics drew sharp pushback from practitioners including Kent Beck and Gergely Orosz. That debate underscored how easily measurement can drift from improvement into surveillance when frameworks are applied without clear intent.

The academic foundation for much of this work, including Nicole Forsgren and colleagues' research published in Accelerate, continues to anchor industry practice. But as DORA notes, the value of any framework lies in how it informs decisions and conversations rather than in the dashboards it produces. Goodhart's law looms over the field: when a measure becomes a target, it tends to lose its usefulness as a measure.

The practical takeaway from DORA's guidance appears to be sequencing. Organizations are encouraged to first articulate the outcomes they care about — faster delivery, higher quality, healthier teams, more predictable planning — and then select the framework or combination of frameworks that best illuminates progress toward those outcomes. Tooling and instrumentation come last, not first.

This perspective is likely to resonate with engineering leaders who have invested in measurement programs only to find that data alone does not drive change. By foregrounding goals over metrics, DORA seems to be nudging the conversation back toward the qualitative judgment and team-level reflection that have always accompanied effective performance improvement, even in technically mature organizations.