Twelve Ways to Improve Your Estimating Accuracy (Part 1)
by Adele Sommers
Did you know that estimating is an essential tool for managing project risks? Whenever we can determine our schedule and budget requirements with reasonable accuracy, it reduces our risk of running out of time, resources, and funding during a project. Yet with all of the emphasis we place on creating accurate estimates and bids, we still seem to have difficulty developing realistic predictions of our time and effort.
If we look carefully at the evidence, I believe we'll find three basic, underlying clues to the reasons for our challenges with estimating:
-
 The presence of hidden or unknown variables that are difficult or impossible to anticipate, and sometimes even more difficult to resolve.
- Our frequently idealistic views of our own capabilities. We tend to imagine that we can accomplish much more than is possible in the time allocated.
- A strong human desire to please other people by telling them what they want to hear. (After all, who wants to be the bearer of bad news?)
This article discusses the first five of a series of tips that will help you and your team produce more accurate calculations.
1. Maintain an estimating and "actual hours" database.
Over time, it's ideal to collect historical data on all of your project undertakings, whether or not you are the person leading the projects. The data you collect will reflect the time that your effort actually took compared to the hours you originally estimated. Try to accumulate numbers for fairly well-defined "task buckets," such as for teleconferences and meetings, research, design, usability testing, various types of content development, reviews, beta testing, and so forth.
 This technique captures proof of your own experience, and produces more realistic estimates that will be tough for others to dispute -- especially when you tell them an estimate is based on your records of actual hours for similar tasks on various other projects.
This method also provides the data you need to develop a realistic buffer, or "fudge factor," to adjust initial estimates that may be way too low or too high. If you tend to estimate on the low side, for example, you can multiply your figures by a factor supported by historical data (such as a multiplier of 200% or even more). These calculations help correct the differences between your initial estimates and the more realistic time required, based on past experience.
2. Create project planning documents.
A project plan is a coordinating tool, planning aid, and communication device that helps you, your organization, team, clients, service suppliers, and contractors define the crucial aspects of your project or program.
By documenting your answers to the 17 topics on the planning checklist, you and your team can think through many of the high-level considerations that can affect your bid or estimate. These include identifying exactly what you will be delivering in the way of products, services, or both.
You might also need to prepare one or more specifications to describe the functions and features required in a final product, service, or custom solution.
3. Perform a detailed task analysis.
For each deliverable you plan to produce, consider providing a fairly detailed, step-by-step breakdown of exactly how you're planning to accomplish each task of the proposed effort.
A bottom-up task analysis helps ensure that you've accounted for every action that someone will perform. It's quite possible to overlook many details when you're viewing things only from the top down. Task analyses also help people unfamiliar with the work understand the steps involved, so they will be less likely to challenge your estimate.
As you perform the analysis, assign the number of hours you think each step will require. If you are not directly familiar with the work, try either to use "actual hours" data, or spend time interviewing or observing a task performer in action. Otherwise, you could make incorrect assumptions about the task complexity. And be sure to include plenty of time for reviews, revisions, and rounds of testing.
4. Use a "complexity factor" to compare two sets of project tasks.
If you are using a previous project as a basis of comparison for your next one, a complexity factor (which is different from the "fudge factor" discussed above) can reinforce the rationale for your estimate.
 A complexity factor indicates which aspects of your new project seem simpler and which aspects seem more complex than those of a previous project. For example:
- "The next project appears about 300% (three times) more complex than the previous one because it requires six new modules, eight new tutorials, and redesigned navigation."
- "Project X will probably take about 1.5 times longer to complete than Project Y because there will be at least 50% more content to create, review, test, and deliver."
5. Use more than one method to arrive at an estimate.
You can "triangulate" by creating three types of estimates, and then see how closely they match one another. For example, you can:
- Apply a "fudge factor" to an initial estimate (as explained in #1, above).
- Prepare a detailed task breakdown (as explained in #3, above).
- Use a complexity factor to compare two projects (as explained in #4, above).
 If the results from each of these methods turn out to be exactly the same, it would give you a very high degree of confidence in your predictions. Realistically, though, the numbers will probably vary somewhat. If so, the next best approach could be to locate the midpoint in the range of these sets of numbers.
For example, if the first method says the effort will require 10 hours, the second method says the effort will require 20 hours, and the third method says the effort will require 30 hours, the midpoint, or average, would be 20 hours.
Just keep in mind that the smaller the range you are dealing with, the less variability exists, and the greater the confidence you can have in the estimate. A range from 10 to 30 hours is fairly large, however. If you found this much variability when using multiple estimating methods, you would probably want to include a cautionary note when you submit your calculations.
In conclusion, maintaining an "actual hours" database and using project plans, detailed task breakdowns, complexity factors, and more than one estimating method are five ways you can increase your estimating accuracy. Part 2 continues the series with an additional set of tips.
Copyright 2011 Adele Sommers
|
It's exciting to imagine how many great new beginnings the remaining months of 2011 will bring! At this very moment, if you're anything like me, you are probably incubating several novel ideas that you would love to hatch this year.
When my clients or colleagues invariably ask, "How long do you think this effort might take?" I usually experience a knee-jerk reaction. Instinctively, a part of my brain that once excelled at solving math problems on timed quizzes goes into overdrive. "I know the answer!" it screams.
Want to organize your project collaborators without resorting to expensive desktop software?
