Business Analysis: Functional & Non-Functional Requirements

Learn Requirements Decomposition / Drill-Down Techniques, Then Judge If AI Output is Right, Wrong, or Just Sounds Good

Functional and Non-functional Requirements Make or Break Your Project
The Standish Group has tracked IT project outcomes for over thirty years. The numbers have barely moved: only 31% of IT projects succeed, and incomplete or poorly defined requirements consistently rank as the leading cause of failure.

That is not a technology problem. It has never been a technology problem. Generative AI does not fix it. It amplifies it. AI tools will decompose your business requirements into functional and non-functional specifications quickly and confidently. They have no way of knowing whether what they produced is complete, correct, or quietly missing something your solution cannot be built without. That judgment still belongs to a human. This course is how you develop judgment.

Why the Sequence of the Sections Is Critical

Sections 2 and 3 teach you our proven decomposition technique through exercises and real-world examples. You will work through the process of decomposing requirement statements and user stories into functional requirements (what the solution must do), data requirements (what it must handle), and non-functional requirements (performance thresholds, usability standards, reliability constraints, and the business rules that govern behavior). Angela and I developed and refined these techniques across hundreds of projects on multiple continents over four decades. They are field-tested, not theoretical, and that distinction tends to matter when the project is real and the stakes are not hypothetical.

NEW: To support your NFR work, the course now includes "The Complete “Ilities” Guide: 121 NFR Quality Attributes, Testing Criteria, and AI Integration Standards, Simply Put!" as a downloadable resource. It catalogs 121 NFR quality attributes with plain-language descriptions, testable acceptance criteria, and AI integration standards. Most teams work from a list of eight or ten attributes. Now you will have 121. Save it somewhere safe. You will want it open during the AI demos in Section 4.

Once you have mastered the fundamentals, section 4 shows you how to bring generative AI into the workflow. You use the AI tool of your choice to suggest the decomposed functional and non-functional requirements (quickly, at scale) and then you deliver human-in-the-loop judgment to evaluate what came back. In practice that means comparing AI suggestions against your actual business context and flagging what is missing or wrong. AI tools generate hallucinations with the same confident tone they use for everything else. Your job is to catch them before they enter development.

Why the Foundation Matters

AI tools have no way of knowing your organization's specific constraints, your stakeholders' unstated assumptions, or the regulatory environment your solution has to work within. A practitioner with solid decomposition skills can evaluate AI output in minutes and catch what is missing. Without that foundation, you are essentially approving whatever the tool decided to say — and misinterpretations tend to surface at the worst possible time (e.g., in system testing or, worse yet, in production).

NOTE: This course is tool-agnostic. Specific AI products come and go. The AI prompting skills taught here transfer to whatever tool our organization happens to be using.

What the Exercises Are For

To maximize the learning effect, the course includes assignments for each technique. They are not tests in the conventional sense. Their purpose is to demonstrate each technique in a context that is closer to real life than an explanation alone can achieve. If you work through them, the techniques will stick. If you skip them, the techniques will still make sense – they will just take a little longer to become second nature on the job.