ISO 25018:2026 — Software Quality Measurement Reference Model

ISO 25018:2026 provides a measurement reference model and practical guidance for planning, implementing, and evaluating software quality measurement programs. As part of the ISO/IEC 25000 SQuaRE series, it focuses on the 2502n measurement division and extends the foundational concepts of ISO 25020 (Measurement Reference Model and Guide) with concrete measurement processes, quality measure definitions, and reporting templates tailored for modern software development environments.

The Role of ISO 25018 in the SQuaRE Measurement Framework

The SQuaRE measurement framework, as defined in ISO 25020, establishes a three-layer architecture: quality model elements at the top, quality measures in the middle, and measurement functions at the base. ISO 25018:2026 operationalizes this architecture by providing ready-to-use measure definitions for each quality sub-characteristic defined in ISO 25010. For each measure, the standard specifies the measurement method, measurement scale, unit of measurement, and interpretation guidelines.

The 2026 revision introduces significant updates to address emerging quality concerns in cloud-native and AI-assisted software systems. New measures cover aspects such as model accuracy drift, inference latency variability, and container orchestration reliability. These additions reflect the expanding scope of software quality in the age of machine learning operations (MLOps) and platform engineering.

Engineering Design Insights for Measurement Program Design

A central engineering insight from ISO 25018:2026 is the concept of measurement purpose alignment. The standard emphasizes that every quality measure must be explicitly tied to a decision-making context. A metric collected without a clear decision purpose is organizational waste. The standard therefore categorizes measures into three decision levels: strategic (for portfolio-level quality investment decisions), tactical (for project-level quality control), and operational (for team-level quality improvement).

The standard also introduces the measurement information model, which defines how raw data is transformed into actionable information. The model specifies: (1) base measures captured directly from the software artifact or process, (2) derived measures computed from two or more base measures, and (3) indicators that interpret derived measures against predefined decision criteria. This model prevents a common anti-pattern where teams collect raw data but fail to convert it into decision-relevant information.

Another key design insight is the measurement repeatability and reproducibility (R&R) requirement. ISO 25018:2026 mandates that all quality measures must include a statement of expected measurement error under defined conditions. For automated measures (e.g., static analysis results), the error derives from tool configuration variability. For manual measures (e.g., usability expert ratings), inter-rater reliability must be documented. Without explicit R&R documentation, measurement results cannot be compared across time periods or organizational boundaries.

Practical Implementation: Deploying a Measurement Program

Implementing a measurement program based on ISO 25018 involves four phases. First, planning — identify the quality goals, select relevant quality measures from the standard’s catalog, and define data collection mechanisms. Second, deployment — integrate measurement probes into the CI/CD pipeline, configure dashboards, and train teams on data collection procedures. Third, operation — collect base measures, compute derived measures, and generate indicator reports at predefined cadences. Fourth, evaluation — assess the effectiveness of the measurement program itself, identifying measures that no longer serve a decision purpose and retiring them.

An important practical consideration in measurement programs is data quality assurance. ISO 25018:2026 emphasizes that measurement data must be validated before it is used for decision-making. Common data quality issues include missing values, outlier measurements from instrument malfunctions, and systematic biases from non-representative sampling. The standard provides guidelines for data validation procedures, including range checks, consistency checks across correlated measures, and statistical outlier detection. Organizations should implement automated data quality checks within their measurement pipelines to flag suspect data points before they propagate into indicator calculations.

A practical technique recommended by ISO 25018 for measurement program design is the measurement objectives tree. This technique starts with high-level organizational quality goals and decomposes them recursively into measurable sub-goals, ultimately arriving at specific base measures. Each node in the tree documents the rationale linking lower-level measures to higher-level objectives, creating an explicit traceability structure. The measurement objectives tree serves as a communication tool between quality engineers, project managers, and business stakeholders, making the rationale for each collected measure transparent and auditable. Organizations using this technique report significantly higher stakeholder buy-in for measurement programs compared to those that select measures on an ad-hoc basis.