How to Measure Anything! ... Or Actually Just Quality Attributes
May 2018 • Presentation
Many quality attributes can be hard to measure. This talk applies lessons from Hubbard's How to Measure Anything to quantify even the trickiest QAs.
Software Engineering Institute
Quality attributes are at the heart of software architecture, yet without concrete metrics they can be abstract and have less business value. Some quality attributes, such as availability, have well-understood and straightforward ways to measure how well a system is doing. Others—such as maintainability, security, or interoperability—are arguably more vague and hard for which to find objective metrics.
This creates a problem. If we don't measure them, how do we know if we're getting better or, worse, going backward? How do we convince our stakeholders to invest time and money in quality? How can we justify the tradeoffs we make?
In 2007, Douglas W. Hubbard released How to Measure Anything: Finding the Value of Intangibles in Business. It posits that anything can be measured, including topics that many have previously considered immeasurable, such as customer satisfaction, organizational flexibility, and technology risk.
This talk brings Hubbard's approach to software architecture. The underlying strategies can be applied to quality attributes: real-world examples will illustrate how any quality attribute can be measured in different contexts and team environments. The talk also covers how to have conversations with stakeholders about quality attributes and make them a first-class concern in the software development process.