Home >

What is a Project Estimate?

In order to run the project we must know how long things take, how much they will cost, and what kind of resources will be required. The only way we can get this data is by doing good estimates. Without good estimates we have no way of knowing where we are at any point in the project, and we have no way of predicting how much the project will cost or how long it will take to do it.

An estimate is the determination of a likely quantitative result. There are two major things that we estimate in a project; one is the cost of the project or the money that will have to be spent to produce it. The second is the time that the project will take to be completed. Whenever we are doing project estimates, we will not only be estimating the cost of doing the work but also the time that it will take to complete it.

Tell me more

There are many pitfalls in producing a good estimate for a project. The deliverables may not all be identified, stakeholders change their minds, project team members may be optimistic or pessimistic, time may be limited, and so forth. If the project is poorly defined, there is not much of a possibility that the cost and schedule estimates are going to come out anywhere near what the actual cost and schedule time for the project will be. Optimistic schedules can cause problems in estimating as well. Stakeholders or management frequently shorten schedules without adding budget to the project. Generally we can look for increases in cost when schedules are shortened. An inaccurate work breakdown structure causes work tasks to be missed. When the individual estimates for the tasks are added up to make a bottom-up, definitive estimate for the project, missed work tasks cause underestimation. Understating risks underestimates our cost and schedule estimates as well. Risks that are not identified and identified risks that have the wrong value for their estimated probability or impact cause management reserves and contingency budgets to be misstated. Cost inflation and failure to include appropriate overheads cause erroneous estimates. It is important to recognize wage and price increases that will occur during the project and adjust estimates accordingly.

Whatever estimating method is used at the appropriate time in the schedule, a range of possible values for the estimate should be given. This range of values should be accompanied by an estimate of the probability that the actual cost will fall within the given range of values. An estimate of $1,000 does not mean nearly as much as the estimate $995 to $1,005 with a 95 percent probability that the actual cost will be in this range. Too many estimates are given in a single value. When the range of values is not known, much needed information is missing and bad decisions can easily be made.

Suppose that in the above example, we gave the manager the cost estimate of $1,000. Suppose that manager had to make a decision as to whether or not to market a new product for the company. If the marketing department says that they can sell the product at a profit with a price of $1,000, should the manager go ahead with the product?

All estimates really have a range of values. There is very little chance that an absolute estimate like our estimate of $1,000 will truly be the actual cost of the product. It will probably be different from that exact value by some amount. It is reasonable to say that all estimates should be given in terms of a range of values.

Suppose that our manager says that the estimate for the product is $1,000, and the sales department says that this will be profitable. The manager decides to go ahead with the product introduction.

Suppose we had told the manager that the product could be built for $995 to $1,005 and that we had a 95 percent probability that the true cost of making the product would fall in that range of values. The manager would probably decide to go ahead with the product and recommend that the selling price be $1,005.

If, on the other hand, we had given her an estimate of $900 to $1,100 and said that we have a 90 percent probability that the actual cost will fall within this range of values, the manager might feel that trying to sell the product at $1,100 could be a very small market penetration and decide not to introduce the new product.

Of course, what this really tells our manager is that the estimate needs to be studied more and the accuracy needs to be improved.

A bottom-up estimate is one in which the cost estimates are independently made for individual, small details of the project. These individual cost estimates are then added together or rolled up to make the cost estimates for the entire project. They can also be added up to create estimates for subprojects within the project. A top-down estimate is one in which the entire project is estimated as a whole, and this value is divided into the component subprojects of the project.

We can think of a bottom-up estimate in terms of the work breakdown structure. If we were to estimate the cost of each task at the bottom of the work breakdown structure independently and add the individual estimates together, we would have a bottom-up estimate for the project.

Bottom-up estimates for the project are inherently more accurate than top-down estimates. This is intuitively true. It is mathematically true as well. When a large number of small details are individually estimated and added together, there is a chance that the individual detailed estimates will be either high or low. That is, some of the individual details will be underestimated, and some will be overestimated. When we add them together, some of the overestimates will cancel out some of the underestimates. The result is that the total project estimate’s accuracy will improve simply by creating more detailed estimates. Of course, since small details are less likely to have forgotten details within them, overall estimates will improve considerably more.

Top-down project estimates are appropriate when we are doing estimates early in the project or at times when relatively inaccurate estimates are acceptable. These estimates are called rough order of magnitude or budget estimates. When it comes time to commit serious money to the project, bottom-up estimates, often referred to as definitive estimates, should be used.

Parametric and analogous estimating methods are used as types of top-down estimates. With the analogous estimating technique we identify another project or subproject that will be used as the basis of the estimate, and we scale it up or down to match the project or subproject we are trying to estimate. If the actual cost of the basis project was collected accurately and if there is a great deal of similarity between the basis project and the project being estimated, these estimates can be quite accurate. For example, a contractor is estimating a project to build fifteen houses. In the past the contractor has built projects with up to ten houses. He could estimate the cost of the new project by scaling up the ten-house project by 1.5.

Parametric estimates are based on some parametric relationship between cost and the parameter that can be measured for the project. In a parametric estimate we use some measurable parameter that changes in the same way that cost does. We then find an adjusting factor that will relate the parameter to the cost of the item being estimated. For example, suppose we wanted to estimate the cost of driving to California. If we know from past experience that gasoline, tires, oil, and wear and tear on the car cost $0.35 per mile and if the distance to California is 2,000 miles, we could estimate the cost of driving to California at $700. Building contractors frequently quote the cost of building residential houses as being $70 per square foot. A 5,000-square-foot house could thus be estimated at $350,000.

Generally speaking, order of magnitude estimates have an accuracy of about −25 percent to +75 percent; budget estimates have an accuracy of about −10 percent to +25 percent, and definitive estimates have an accuracy of about −5 percent to +10 percent. You may ask why the range of values here is lopsided. This is simply realistic. If we estimate the cost of something to be $50 it would be impossible for us to be more than $50 overestimated but it is quite possible for us to be $50 or even more underestimated.