Estimating is tricky for project managers who have to balance conflicting pressures from the sponsor, stakeholders and their team:
- The customer or user wants the project done quickly and cheaply.
- You, as PM, want to finish on time and within budget.
- For commitment, the team needs to participate in a process their perceive as fair and not feel like they are sure to fail because their estimate is impossible.
- The estimating technique should yield accurate numbers and some assessment of the accuracy.
- Decision makers need information of the certainty of the project finishing on time
That list of requirements is a tough one for any project estimating process. The only process that meets all those requirements is 3-point estimating, which formerly called PERT (Program Evaluation and Review Technique).
Briefly, 3-point estimating has three-steps. In each, the PM works closely with the people who will be doing the work. The first step is to discuss the deliverable the team member will be accountable for producing. This discussion includes the “good” risks that could cause this task to take less work and the “bad” risks that could cause it to take more work. Second, the PM notes these risks on the work package form that also contains the approach the team member will use. Third, the team member makes three estimates: an optimistic estimate, a pessimistic estimate and a best guess estimate. The PM applies the 3-point formulas (at the end of the article) to those three estimates to come up with the actual data that you will use in the project schedule.
Common Estimating & Risk Issues
Two mindsets often plague the estimating process:
- Executives often believe that projects have no risks that affect duration or budget.
- Team members think that padding their estimates will protect them from blame.
Both of these mindsets are false but they certainly get in the way of accurate estimating.
The 3-point estimating technique deals with both these mindsets. It gives PMs a data to communicate the risk of a work estimate. It also lets everyone stop pretending that task #135 is going to finish in precisely 15 days or that the project will absolutely finish by August 30. Three-point estimating is a straightforward process for developing estimates using just a little bit of statistics. It gives you a tool to address the issue that most projects are launched with less than a 35 % chance of finishing by their promised due date. Because no one talks about that issue, executives think the completion date is 100% guaranteed. It’s only missed when someone goofs off.
The best project managers have risk data for their sponsors. They can document why a project has a 65% chance of finishing by August 30, as an example. These PMs also explain what they can do to increase those odds to 75% or 90% and what it will cost. Those same PMs manage the assignments of their project team members with an understanding that there is risk on each assignment. They use 3-point estimating techniques to get data on the risks.
Three Point Estimating in Detail
The 3-point estimating process starts with a discussion with the team member about the risks inherent in their assignment. You discuss the bad risks that will make their assignment take more work and duration (time). You also discuss the good risks that will cause it to take less work and duration (time). Why should you do this step? Because you need an estimating process that addresses the team member’s legitimate concern that bad things will happen on their assignment and they’ll be blamed for not meeting the completion date. With agreement on the risks in the assignment and work package notes what you will do about them, you go on to the estimates work and duration.
As the name implies, 3-point estimating requires three estimates for each task. That sounds like it will take a lot of work but it takes a matter of minutes. You and the team member develop an optimistic estimate, a pessimistic estimate and a best guess estimate for each task. By developing those three estimates, we get estimates that are more accurate from team members and assess the assignment’s degree of risk and the range of durations.
Before we go on, we need to talk a little bit about risk. When you ask me how long it will take to read this article, I might estimate five minutes. Am I guaranteeing you that no matter what happens I’ll be able to read the whole thing in five minutes? No, what I mean is that 5 minutes is my best guess. That means there is a 50% chance it will take me less than five minutes and a 50% chance it will take me more than five minutes.
However, if you were my project manager asking me for a task estimate, I would be a little hesitant about giving you an estimate in which there was a 50% chance of an overrun. What I would rather give you is an estimate where I’m 90% confident that I can finish in that amount of time or less. As the project manager, you would probably regard that estimate as padded. As the team member, I feel more comfortable with a 90% estimate. Unfortunately, there is no consistency in the amount of padding your team members will use.
You want your team members to leave the estimating process knowing that you considered the fact that things can go wrong on a task assignment. That’s why you identified risks at the beginning of the discussion and documented what you could do about the risks. With that recognition of the risks, we move on to gathering data on the impact those risks could have on the assignment. Using the three estimates enables you to do that. It’s better than having a team member give you a single estimate and play the padding game about how certain that estimate is. The three estimates tell you the variability in the task.
Best Guess, Optimistic and Pessimistic Estimates
Now let’s start the estimating process. Your team member estimates that a task has a best guess estimate of 80 hours of work. That means that 50% of the time it will take more work and 50% of the time it will take less.
Next, the optimistic work estimate is less work than the best guess. The optimistic estimate is low enough that the team member thinks they can get the task done for less than the optimistic estimate only 20% of the time. The task will require more work than the optimistic estimate 80% of the time.
The pessimistic estimate is more work than the best guess. It is not a “disaster” estimate but we want an estimate that’s based on the bad risks that we identified happening. The pessimistic estimate is high enough that the team member thinks they can get the task done for less than the pessimistic estimate 80% of the time. The task will require more work than the pessimistic estimate 20% of the time.
Now let’s dip our toe into the statistics and look at two tasks, Alpha and Beta, and the calculated work estimates we would use at three different level of confidence (* see formulas below).
What we did was take the three estimates and use some simple formulas to calculate the task’s work estimates and calculate the mean and standard deviation. Using standard statistical tables (z-scores from a table of standardized normal deviates); we can take those means and standard deviations and use them to calculate levels of confidence of finishing within the estimate. In other words, for task Alpha we could say that we have a 50% chance of completing the task with less than 54 hours of work. For an 80% confidence level, we would calculate that 69 hours of work would be required. This is the data to use with a client or project sponsor to quantify the cost of higher levels of certainty about a completion date. In the previous example with Alpha, we have to buy an additional 15 hours of work to move from 50% confidence to 80% confidence of getting the task done within the work estimate. The beta is much less risky task than alpha. The mean work estimates are very close but the standard deviations are very different. To move from the 50% level of confidence that is 50 hours on task beta we would need to increase the work estimate to 51 hours. So for task beta higher levels of certainty a relatively inexpensive. Extending these calculations to the entire project is very easy with a spreadsheet such as the one we use in our classes. It gives project managers the ability to discuss the cost of higher levels of certainty. Sponsors always say they want to be 90% confident of finishing on time. When you present them with the cost of that level of certainty, it often is the case that lower levels of confidence would be acceptable.
Using 3-Point Estimates
All of the better project management software packages, such as Microsoft Project®, enable you to use 3-point (PERT) estimates and create a variety of reports that communicate the project’s risks. You can take estimates like those above and calculate the odds of finishing the entire project within various durations. That information is a solid basis for a discussion with the sponsor about the tradeoffs between cost, scope, duration, risk and staffing levels.
To learn these 3-point estimating techniques and the entire estimating process, consider our private, online courses where you work individually with your instructor. They are available by phone, video conference or e-mail whenever you have a question or need help on an assignment. We can also deliver a customized training program at your site for up to 25 people. Call us at 303-596-0000 and speak to an instructor.
*Three point estimating Formulas
Probability level = work= Mean + (z-score for probability)*SD