A Toolkit for Project Time Estimation
- Date: 05 January, 2017
Abstract
Resource management is always an issue in any project, especially when the stakeholders from whom we need time have operational duties to perform. If our requirements team was at our disposal 100 percent, always completed activities on target, and worked a full eight hour day without distraction or a loss of productivity, then estimating time would be simple. In this paper, we explore standard approaches to time estimation, the dangers of multi-tasking, and estimation alternatives, which consider work habits and productivity norms. We also build something new by combining techniques to create a tool that will determine the statistical likelihood of accomplishing a task.
Sample
Time estimation on even the simplest project is not as easy as it seems. Many business analysts (BAs), project managers (PMs), and other project team members, such as subject matter experts (SMEs), customers, users, and other stakeholders work in a functionally organized environment. Functional settings force us to bring project team members together from other areas within the organization.
This organizational paradigm causes many issues, such as competing for resources, self-interested management behavior, poor coordination between projects, over commitment of resources, and a fundamental disregard for best practice project planning techniques. In the case of opportunistic management behavior, many functional organizations allocate resources based on project priority. In such cases, there is an incentive for project sponsors and senior managers to keep priorities high by any means possible. On the other hand, those who already have resources assigned to their projects would want to protect them from poaching. As we may expect, this poor time management behavior leads to a negative effect on project accounting practices.
Organizations often account for costs based on hours spent by team members on projects. In contrast, time devoted to internal activities, such as meetings, is viewed as a non-project expense. As a result, there is a built-in incentive for management to keep as many people as possible working on projects. A side effect of this is the lack of availability of resources for new projects. Moreover, real project costs are never identified and meetings are not tracked.
If our requirements team was at our disposal 100 percent of the time, always completed activities on target, and worked a full eight-hour day without distraction or a loss of productivity, estimating time would be a mere 1:1 ratio.
In this paper, we will understand the real cost of multitasking. Next, we will explore standard approaches to estimation and alternatives to a single-point estimate such as the requirements elicitation, planning, analysis, and collaboration (REPAC) technique. We will also consider how elements such as productivity, cumulative probabilities and probability densities, and Bayesian reasoning can help us account for the uncertainty given in estimates. Lastly, learn how to determine the overall likelihood of task completion by the desired date.
The Problem
Planning project timelines is not absolute. Estimates based on probabilities must always come with a margin of error. Figure 1 illustrates the best practice order of magnitude with respect to project time and money. (This is known as The Cone of Uncertainty). Classic time estimation on projects done using two simple techniques yields a broad range of results. The first method, as seen in Equation 1, assumes that the person giving the estimate can account for the variables needed to provide a thoughtful assessment. Regrettably, this approach has a less than 30 percent success rate (Standish Group, Chaos Report, 2012) and is, ironically, used in most cases—even when the project timelines carry substantial risk.
