When critical subsystems fail, the resulting losses can be catastrophic. In the insurance industry, if premiums are miscalculated, defect costs can reach well over a million dollars. In this session, Mike Keith and Dom Nunley draw on their practical experience with insurance systems testing to provide an overview of combinatorial automation testing for high-risk backend system areas—i.e., features that absolutely must work correctly. They share a process for categorizing requirement risk levels to determine which requirements warrant combinatorial testing. Mike and Dom illustrate various combinatorial testing techniques such as N-FAT, N-Wise, and RANDOM, which can be used to automatically generate test cases. These methods are used to ensure coverage against risk while controlling the number of tests that run.

Google pioneer Alberto Savoia offered this sage advice: Build the right "it" before you build It right. But few software companies take the time to define, much less build, the right "it." The problem starts with a poor requirements definition process. In this session, join Kathryn Campbell as she examines the five most common mistakes that software companies make during requirements definition—and how to avoid them. First Kathryn defines thinking too small as a huge problem and shows you how to broaden your perspectives. Next, she exposes being stuck in the past, with legacy systems maintaining too much control of our innovation. The third mistake is assuming too much about your customers. Kathryn shares guerrilla techniques for gathering rapid, inexpensive customer feedback at every stage of your requirements and design process.

User stories and their big brothers, epics, are an excellent way to describe requirements for a software system. They act as stakes in the ground to keep track of what the system needs to do, the type of user most interested in each feature, and the reason the requirement...

Test plans are essential for communicating intent and requirements for testing efforts, but excessive documentation creates confusion—or just goes unread. Try the 5W2H method. The name comes from the seven questions you ask: why, what, where, when, who, how, and how much. That's all you need to provide valuable feedback and develop a sufficient plan of action.