An analysis of nearly 100 organizations and individuals reveals where the industry excels, where it struggles, and the critical factors demonstrated by the best adopters of CD. Could these findings help you identify your CD strengths and weaknesses, and suggest ways you can improve?
By Dave Farley
Summary of Key Findings
The organizations involved were confident in applying a little over half of the essential CD practices.
Most of these organizations have grasped the cultural and technical challenges and perform well in respect of: Pipeline Integrity, Autonomous Teams and Process Discipline.
The main gaps are: Trunk-based development, Test Automation, and end-to-end Pipeline automation.
About 10% organizations are ‘experts’; achieving comprehensive CD implementation. What is it they get right?
There is a correlation between Trunk-based development and Test Automation, with the best performing organizations doing well on both, while developing or beginning practitioners are unable yet to adopt either.
Based on these findings, we go on to recommend a three-stage approach for organizations seeking to advance to the level achieved by the top 10%.
You can DOWNLOAD the complete paper here, or continue reading… |
The evidence and data say that Continuous Delivery is the most effective way to create software in terms of speed and quality[1]. Continuous Delivery means working so that your software is always ready to release. It does not require you to release continuously, but rather that you maintain your software in a releasable state, and so are able to release it whenever makes sense - when you, or your customers, choose. Achieving this involves several technical practices, team behaviours and organizational factors that need to be aligned.
From decades of software engineering experience and working with software development teams around the world, I have identified 14 attributes that I see as essential determinants of Continuous Delivery.
Through a Continuous Delivery Assessment and Analysis Tool, used in my consultancy and training, I have collected information from nearly 100 organizations and individuals over the last 2 years, about which of these CD techniques they are good at, what they need to improve on, and what it takes to do Continuous Delivery really well.
Analysis of this data paints a nuanced picture of an industry in transition, one that has (for this group) largely solved the cultural and process challenges of CD, but remains stymied by some foundational, technical practices.
If you’d like to learn more about the 14 essential practices of Continuous Delivery, and try the CD Assessment and Analysis Tool, you can join the CD.Training programme here, with a special 30% discount. |
I provide tools to guide people through an assessment of their CD practices as part of my consultancy work and Continuous Delivery training courses. So the contributors were familiar with the core ideas behind the 14 principles being investigated, and so better informed than the general population and presumably self-selecting in the sense of being pre-disposed to wanting to learn more about CD.
The assessment tool helps software development teams identify where their strengths are, find any gaps and explore opportunities to improve. They can also compare how well they are doing in relation to other CD practitioners.
Overall average maturity sits at 53.2%, with a small but encouraging upward trend of 5% improvement over the past two years. This suggests that organizations are genuinely investing in CD practices and seeing results, not merely paying lip service to the concept.
However, the distribution reveals a mature market with clear segmentation:
This distribution suggests a maturing market where the "early adopter" phase has largely passed, and we're now seeing broad-based adoption across different organizational contexts.
The Cultural Victory: What the Industry Gets Right
Perhaps the most striking finding is that organizations have largely solved the cultural and process challenges that dominated CD discussions a decade ago. Three practices consistently score above 60%:
Pipeline Integrity (67.9%): Organizations understand that what gets tested should match what goes to production. This represents a fundamental shift from the "it works on my machine" mentality that plagued software development for decades.
Autonomous Team Structure (61.9%): Small, cross-functional teams with end-to-end responsibility have become the norm rather than the exception. This addresses Conway's Law concerns and enables the organizational agility that CD requires.
Process Discipline (61.9%): Changes consistently flow through well-defined pipelines rather than ad-hoc deployment processes. This foundational discipline creates the reliability and repeatability necessary for frequent, safe, deployments.
These results suggest that the industry has internalized many of the cultural lessons of the DevOps and Agile movements. The days of throwing code "over the wall" to operations teams, for this group at least, are largely behind us.
However, these data reveal a stark divide in technical skills that separates the good organizations from the great ones. Three practices consistently lag behind, creating what we term the "technical skills cliff":
Trunk-Based Development (35.0%): Is fundamental to the operation of Continuous Integration, Continuous Delivery is foundationally built on the concept of Continuous Integration, and so CI is core to its effective operation. CI allows us to identify, and verify, the current state of the system. CD builds on this to evaluate the “Releasability” of that current state of the system. So problems here represent the single largest barrier to CD adoption. Despite decades of advocacy, the majority of organizations still rely on long-lived feature branches, creating integration delays, reducing deployment frequency and limiting visibility into the current state of the system.
Test Automation (41.9%): CI & CD depend on fast, efficient, definitive, automated tests that can give us fast feedback on the correctness of our systems, and in particular, their releasability. CD then relies on the pervasive use of high-quality automated tests throughout the development process. The foundation of safe, frequent deployments remains elusive for most organizations. Without comprehensive automated testing, teams cannot confidently deploy to production multiple times per day.
Complete Deployment Pipeline (47.1%): To achieve this fast, definitive feedback on the releasability of our changes, we need the automation in the deployment pipeline to be definitive. If the pipeline says our change is good to go, there is no more work to do prior to release. Many organizations have achieved partial automation but lack the end-to-end pipeline automation necessary for true CD.
Our analysis reveals a troubling correlation: 27% of organizations can neither implement trunk-based development nor effective test automation. This isn't coincidental, in fact these practices are mutually reinforcing. There is a strong correlation between these technical practices, orgs are either good or bad at both, but rarely good at one and bad at the other.
Trunk-based development requires comprehensive automated testing to ensure code quality when multiple developers integrate frequently. Conversely, effective test automation is difficult to maintain across long-lived branches due to lack of clarity about what is the definitive current version of the system.
Organizations stuck in this "correlation trap" face a chicken-and-egg problem: they need automated testing to safely adopt trunk-based development, but their branching strategy makes comprehensive test automation significantly more difficult to achieve.
The 10.7% of organizations achieving expert-level CD implementation provide a template for industry advancement. Their practices show consistent strength across all dimensions, but notably:
These organizations demonstrate that the technical challenges are solvable, but require sustained investment in engineering practices rather than just process improvements.
The largest segment (32.1%) consists of organizations scoring between 41-60% on most practices. This "intermediate plateau" represents a critical challenge for the industry. These organizations have moved beyond basic DevOps practices but lack the technical depth to achieve true CD excellence.
Analysis of their patterns reveals a common profile: strong in process and culture, weak in advanced technical practices. They've implemented the "easy" parts of CD (team structure, basic pipelines, process discipline) but struggle with the technical engineering practices that enable rapid, safe deployment.
This suggests that incremental improvement may not be sufficient. Organizations need decisive action to break through the technical barriers rather than gradual evolution of existing practices.
Based on these findings, we recommend a three-stage approach for organizations seeking to advance beyond the intermediate plateau:
Stage 1: Test Automation Foundation Before attempting trunk-based development, organizations must establish effective, comprehensive, automated testing. This includes not just unit tests, but integration tests, contract tests, and automated acceptance tests. The investment in test infrastructure pays dividends across all other CD practices.
Stage 2: Trunk-Based Development With solid test automation in place, organizations can more safely transition to trunk-based development. We recommend starting by limiting the maximum duration of branches, start by adopting short-lived branches (< 2 days) as a stepping stone before moving to true trunk-based development.
Stage 3: Complete Pipeline Automation The final stage involves eliminating all manual steps in the deployment pipeline. This includes automated environment provisioning, database migrations, and rollback capabilities. This will improve the stability of the testing and release process and so enable and facilitate higher quality testing and deployment.This is made MUCH easier when Trunk Based Development is in-place, because TBD in this mode fixes the dependency management problems that otherwise so often arise.
Industry Implications: The Technical Skills Gap
These findings, if more generally true, have profound implications for the software industry. While we've made significant inroads into the cultural and process challenges of CD, we face a continuing technical skills gap. The practices that separate good organizations from great ones (trunk-based development, comprehensive test automation, and complete pipeline automation) require deep technical expertise that many organizations lack.
This suggests several industry-wide challenges:
Education and Training: Computer science curricula and bootcamps must emphasize these advanced engineering practices alongside basic programming skills.
Tooling Evolution: The current generation of CD tools assumes technical sophistication that many organizations don't possess. We need tools that make advanced practices more accessible.
Organizational Investment: Companies must recognize that achieving CD excellence requires sustained investment in technical capabilities, not just process improvements.
The good news is that technical problems like these are usually easier to overcome than organizational and cultural barriers.
These data tell a story of an industry in transition. We've moved beyond the early adopter phase and solved the fundamental cultural challenges that once seemed insurmountable. However, we've reached a new plateau defined by technical rather than cultural barriers.
The 5% year-over-year improvement suggests that progress is possible, but the concentration of organizations stuck in the intermediate range indicates that breakthrough requires more than only incremental improvement. Organizations must make decisive investments in technical practices to achieve CD excellence. We need technologists to approach these changes with a clearer vision and a more effective mindset. The CD philosophy grounded in Modern Software Engineering techniques facilitates this focus and this change of perspective.
For the industry as a whole, this represents both a challenge and an opportunity. The organizations that successfully navigate the technical skills cliff will gain significant competitive advantages through faster delivery, higher quality, and greater responsiveness to market changes.
The question isn't whether Continuous Delivery works, the expert practitioners prove that it does. The question is whether the industry will make the technical investments necessary to move beyond the intermediate plateau and achieve the promise of truly continuous delivery.
If you’d like to learn more about the 14 essential practices of Continuous Delivery, and try the CD Assessment and Analysis Tool you can join the CD.Training programme here, with a special 30% discount. |
You can DOWNLOAD the complete paper here, to read and reflect on later. |
Dave Farley is a software developer, trainer and consultant, creator of the Continuous Delivery and Modern Software Engineering YouTube channels, and author of "Continuous Delivery", "Modern Software Engineering", “Continuous Delivery Pipelines” and "The Software Developers’ Guidebook”.
[1] See “Accelerate, The Science of Lean Software and DevOps” by Nicole Fosgren, Jez Humble & Gene Kim
