Bang for the Buck Test Automation
Elisabeth Hendrickson · 2001
Page 1 of 7 Bang for the Buck Test Automation Elisabeth Hendrickson January 2001 You’ve heard the horror stories. Hundreds of thousands of dollars spent on test automation that found 2 bugs. Thousands of person-hours spent writing automated tests that became useless as soon as the interface changed. Test automation tools that became shelf-ware within a few weeks of purchase because the organization wasn’t really ready to automate. In 1994, I worked on a test automation project that failed spectacularly. The test manager asked us to “automate everything.” A few weeks into the project, the manager asked me, “so how long before we see some benefits?” “It will take us about 2 years to automate everything,” I replied, giving him the full estimate that my team and I had worked out. “2 years!?!” “Yup.” The manager’s face clouded as he waved his arms. “That’s not acceptable. I want you to think about Bang for the Buck!” I did. Since then, I’ve used test automation as a part of an overall test strategy. As a result, I have examples of good test automation stories. A tool that took a day to write found serious bugs that would not have been found otherwise. A script used information about the software under test to generate tests as it executed. Small disposable scripts set up the test environment and saved testers hours. These stories are not about comprehensive press-a-button-and-the-tests-run-overnightwhile-you-sleep test suites. They aren’t about attempting to reach a goal of 100% automated. They’re about writing simple tools to make the process of testing easier. These stories inspired this paper.