The subject of this article is the methods, principles, and models for building a system of metrics and key performance indicators (KPI) for the CI/CD pipeline, considered as a complex organizational and technical management system. Modern software development demands increasing speed and reliability in delivering changes; therefore, continuous integration and delivery (CI/CD) practices are gaining strategic significance. The traditional approach, which focuses solely on the functional operability of the pipeline, proves to be insufficient: the CI/CD pipeline possesses economic value, resource intensity, and reliability characteristics that require systematic measurement and management. The lack of a formalized measurement system leads to a number of management issues: the inability to objectively assess the return on investment in automation, subjectivity in evaluating team performance, the difficulty in identifying bottlenecks, and the risk of local optimization of individual stages to the detriment of the overall effectiveness of the value creation flow. The paper employs a systems-analytical approach that combines cybernetic modeling, principles for constructing balanced scorecards, and causal analysis of the relationships between metrics of the CI/CD pipeline as a management object. The scientific novelty of the work lies in the development of a systems-analytical concept for constructing a measurement system for CI/CD, based on a cybernetic model of the pipeline with clearly defined inputs, outputs, and feedback loops. For the first time, a four-aspect model of effectiveness has been proposed as a unified conceptual framework, encompassing speed, reliability, quality, and cost-effectiveness of the pipeline, unified by five key principles: systemic cohesion, balance and absence of internal contradictions, result orientation, cascading connection to business goals, as well as evolutionary and adaptive capacity. It is shown that a systemic approach requires not merely listing metrics but modeling the causal relationships between them, which allows transforming raw pipeline data into management decisions. In conclusion, practical recommendations for implementation are formulated: phased introduction of the cybernetic model, development of a balanced scorecard, creation of a technological platform for data collection, and formation of an organizational strategy that transforms metrics into a tool for continuous improvement of software development processes.
A. A Kuznetsov (Sun,) studied this question.