Project Management Institute

Estimating techniques

another point of view

img

EXECUTIVE NOTEBOOK

Joan Knutson

Estimating has been described as a stepchild in project management. Project management literature, seminars, and organizational policies and procedures have tended to avoid the subject of estimating. This tendency can be explained, in part, by the discipline-specific nature of estimating. It's difficult to develop a set of universal rules that can be applied to the systems development project, the engineering project, the new product development effort, and other organizational undertakings. There are, however, some com-monsense rules that can be applied, regardless of the substance of the project.

Rule 1: Establish a definition of “best estimate.” Many times an estimator is unaware of the organization's concept of “best estimate.” Some organizations consider the lowest-cost/shortest-duration estimate to be the best estimate. Yet the lowest-cost/shortest-duration estimate may have small probability of being achieved. The best estimate cannot be one that has no possibility of being overrun or late because, under estimating guidelines, it's doubtful that such a project would be approved in the budgeting process; the cost would simply be too high. The best estimate, under most circumstances, is a middle-of-the-road estimate, with an equal probability of being overrun or under-run.

Rule 2: Direct the estimator to provide an estimate that reflects management's priorities. If the project manager knows that the schedule is most critical, ask the estimators to provide minimum time estimates. If keeping the cost down is management's pri-mary goal, ask the estimators for minimum cost estimates. If cost and schedule are of equal importance, ask for an estimate that reflects a balance between time and cost. Rarely, if ever, ask for estimates without providing guidance to the estimators. Without guidance, the result will be an amalgam of whatever is most convenient for each estimator, making it impossible to evaluate the potential for altering the plan through negotiation. If the project manager lacks insight with respect to senior management priorities, he or she should ask for clarification.

Rule 3: Ensure that the estimates are prepared/reviewed by the personnel responsible for performing the work. Develop commitment to the estimates, motivation to meet planned targets, and a sense of ownership of the plan through participation. These attributes will be lacking if the personnel responsible for the execution of the work do not participate in the development of the estimate. This complicates the project manager's job a great deal. It's difficult to hold people accountable for their performance when they had no input in determining the performance measurement criteria.

Rule 4: Estimate at the appropriate level of detail. Estimates rendered at a global level suffer from inaccuracy. Overly detailed estimates cost too much and take too long to develop. Over time, guidelines based upon experience should be developed to indicate what level of detail in task definition yields estimates that are accurate yet cost effective. If projects are always within several percentage points of planned performance, the level of detail in estimating may be too great.

Rule 5: Consider inherent time/cost/ resource use trade-offs in the estimates. Linear thinking can be an enormous problem in the development of estimates. A task that can be performed by one person in 20 days cannot necessarily be performed by two people in ten days, or four people in five days, or 20 people in one day! Few tasks are so perfectly partitionable that they can be performed in any of the combinations above. However, few tasks are non-par-titionable; it's rare to find a task that takes one person 20 days, and which cannot be shortened at all. In reality, most tasks are partitionable with communication, but they follow a non-linear pattern of time/cost/resource use tradeoffs. Thus, one person may do the task in 20 days at a cost of 160 person-hours. Two people may perform the same task in 12 days at a cost of 192 person-hours. The additional 32 person-hours are for communication and coordination. Three people may get the job done in ten days at a cost of 240 person-hours. Communication and coordination time goes up geometrically as the number of people assigned to the task increases. The nature of the task must be considered in the estimating process, as must the formula for the geometric increase in person-hours caused by adding staff.

Rule 6: Factor productivity into the estimates. If the normal workday is eight hours, is it possible to get eight hours of project work completed in a day? The organization's environment does not allow it. Productivity varies from organizational unit to unit, based upon the responsibilities of that unit. Productivity may also be seasonal. Each functional manager or supervisor should be responsible for developing guidelines to be used by their units in preparing estimates. These guidelines ought to be reviewed periodically to determine their continuing relevance.

Rule 7: Do not overload the task! In scheduling critical undertakings there is a tendency to apply additional resources to the effort, to the point at which it is greatly overstaffed. Not only is this a waste of funds and human resources, it may also lengthen the period of performance of the project. Too many resources can get in each other's way, impede progress, and cause accidents, which leads to rework. While some level of effort above the optimum team size may be appropriate, excessive effort should be avoided.

When these rules are applied, together with discipline-specific methods of deriving the estimate, an improvement in the quality of the estimates rendered can be expected. ■

Joan Knutson is president and founder of Project Mentors, a San Francisco-based project management consulting and training firm.

This material has been reproduced with the permission of the copyright owner. Unauthorized reproduction of this material is strictly prohibited. For permission to reproduce this material, please contact PMI.

PM Network • July 1996

Advertisement

Advertisement

Related Content

Advertisement