In the days of large waterfall projects, organizations made the assumption that a software budget was allocated one-third per major category: analysis and design, develop, test. This was the rule of thumb that was used to generate a high-level estimate (HLE). This rule of thumb was great when it was applied evenly to all three categories. However, what usually happened on software projects is that the first two categories needed more time and it inevitably came at the expense of testing in an effort to stay on budget.