Software Estimation Without Guessing (pdf)

Description

In a world where most developers hate estimation and most managers fear disappointment with the results, there is hope for both. It requires giving up some widely held misconceptions. Let go of the notion that “an estimate is an estimate” and estimate for the particular need you, and your organization, have. Realize that estimates have a limited shelf-life, and reestimate frequently if it’s important. When reality differs from your estimate, don’t lament; mine that disappointment for the gold that can be the longer-term jackpot.

Estimate in comparison to past experience, by modeling the work mathematically, or a hybrid of both. Learn strategies for effective decomposition of work and aspects of the work that likely affect your estimates. Hedge your bets by comparing the results of different approaches. Find out what to do when an estimate proves wrong. And they will. They’re estimates, after all. You’ll discover that you can use estimates to warn you of danger so you can take appropriate action in time. Learn some crucial techniques to understand and communicate with those who need to understand.

Estimating software development often produces more angst than value, but it doesn’t have to. Identify the needs behind estimate requests and determine how to meet those needs simply and easily. Choose estimation techniques based on current needs and available information, gaining benefit while reducing cost and effort. Detect bad assumptions that might sink your project if you don’t adjust your plans. Discover what to do when an estimate is wrong, how to recover, and how to use that knowledge for future planning. Learn to communicate about estimates in a healthy and productive way, maximizing advantage to the organization and minimizing damage to the people.