TRAINING IS A MAJOR ACTIVITY of many types of projects, especially those that develop and deploy new business processes, computer systems, or software applications. For example, implementing a new order entry system can change the process of accessing customers' information, entering orders, and tracking product sales. Employees performing the ordering process must learn new methods and procedures, new user interfaces, and new ways to access information. Training activities should include hands-on exercises that will allow employees to productively use the new system. But how many exercises are required? How much training time is needed? And how productive will the employee be at the completion of training? You can use the learning curve concept to answer these questions and help design an effective training course.
The basis for the learning curve concept is “practice makes perfect.” The more you repeat a process, the more efficient you become and the less time it takes to perform. Businesses use learning curves to forecast costs, price, or work hours relative to the number of units produced (output) over time. In the article, “Measuring Learning Costs” (Management Accounting, August 1994), Ken M. Boze wrote, “The 80 percent learning curve is standard for many activities and can serve as an average for forecasting learning costs.” The same applies to forecasting learning time. When applying the learning curve concept to training design, you can estimate the number of training exercises needed to gain a desired productivity improvement level and the total time required to complete the exercises. Let's look at how the learning curve works.
Exhibit 1 is an example of an 80 percent learning curve. The horizontal axis shows the cumulative number of completed exercises (output), and the vertical axis shows the expected time to complete each exercise. In this example, it takes 10.00 minutes for a person to successfully complete the first exercise. When the output doubles to two completed exercises, the expected time falls from 10.00 minutes to 8.00 minutes, 80 percent of the first exercise (or a 20 percent decrease in time). When the output doubles again, to four completed exercises, the expected time falls to 6.40 minutes, 80 percent of the second exercise. Each time the number of completed exercises doubles, there is a 20 percent decrease in time.
Notice that the slope of the learning curve is initially steep, then begins to flatten. This is because a person learns quickly at first, but with more and more practice, becomes proficient at a task. Productivity, in this case the time it takes to successfully complete each training exercise, improves rapidly with the first few exercises, then tapers off as the individual gains proficiency. Therefore, learning happens at a decreasing rate.
So, how many exercises are required for an individual to become sufficiently productive? According to Boze, “The data tell us training programs should contain at least eight to 10 exercises for each activity, which should move the individual well down the learning curve.” You can see why by looking at Exhibit 2. Column A shows the cumulative number of completed exercises. Column B shows the percentage of expected productivity increase from the previous exercise. Column C shows the cumulative percentage of expected productivity improvement from the first completed exercise. Note that percentages in Columns B and C apply to any 80 percent learning curve. Looking at Column B, notice how the rate of productivity improvement decreases with each completed exercise. The second exercise shows a 20 percent productivity increase from the first, but the 20th completed exercise shows only a 1.6 percent productivity increase from the 19th. Column C shows the total effect on productivity. For example, it takes only 10 completed exercises to gain a 52.4 percent total productivity increase, but it takes 10 more exercises, for a total of 20, to gain only an additional 9.5 percent increase to 61.9 percent. This means that the benefits gained from each additional exercise after 10 may not be worth the additional time and cost to your project.
Exhibit 1. An example of an 80 percent learning curve that measures time, in minutes, to complete each training exercise to the cumulative number of completed exercises. Time to complete each exercise decreases as the number of completed exercises increases.
Exhibit 2. Use this table to identify the percentage of productivity improvement with an 80 percent learning curve. (Adapted from “Measuring Learning Costs,” by Ken M. Boze, Management Accounting, August 1994.)
Exhibit 3. Use this table as a tool to estimate the training time required to achieve a target productivity improvement level for each training activity.
The complexity of the activity may determine whether you include more than 10 or less than eight exercises. For example, from a data entry perspective, creating an initial order for a new customer is usually more complex than canceling an existing order. Use Exhibit 2 as a guide for choosing a desired productivity improvement level for each activity. For establishing a new order, you may want to include 12 data entry exercises to achieve a 55.1 percent total productivity improvement from the first exercise. For canceling an order, six exercises for a 43.8 percent total productivity improvement may suffice. This gives you the flexibility to vary the number of exercises to achieve a target productivity improvement level for each activity.
How much training time is required? You can estimate the total time to complete each person's training by adding together the expected minutes to complete the total number of exercises for each training activity. Exhibit 3 shows the 80 percent learning curve as it applies to three training activities for our new order entry system: establishing a new order, changing an order, and canceling an order. Column A shows the cumulative number of completed exercises. Column B shows the learning curve factor for each exercise. Note that the learning curve factors shown are the same for any 80 percent learning curve.
Before you apply the learning curve factors you will need to estimate the expected time to complete the first exercise. You can do this by using parametric (historical and statistical) methods or other estimating tools you have available. Once you've determined the expected time to complete the first exercise, multiply it by the learning curve factor for each subsequent exercise to get the expected number of minutes. For example, Column C shows an estimated time of 15.00 minutes to complete the first “new order” exercise. The second exercise is expected to take 12.00 minutes, .800 x 15.00. Exercise three is expected to take 10.53 minutes, .702 x 15.00, and so on. The same logic applies to Columns D and E, changing orders and canceling orders, respectively. The totals at the bottom of each column reflect 12 exercises for entering new orders, nine exercises for changing orders, and six exercises for canceling orders. Column F shows the total time it takes to complete all 27 exercises, 192.57 minutes (or 3 hours, 13 minutes). With startup activities and breaks, you can plan on a four-hour training class.
At this point you've estimated the number of exercises and the time required for individuals to achieve a certain productivity improvement level at the end of their training. How do you give your students the best chance at continuing to improve in practice? Timing! According to Ken Boze, “Added time between training and practical application shifts the learning curve up, as does time away from the job. Therefore, it is important that users apply what they learn quickly and get plenty of practice. Training should be timed with this in mind.” Ensure that training sessions and deployment of the new system are closely linked in your project plan. To take full advantage of the learning curve, employees should move from the classroom to the workplace in the shortest time possible.
REMEMBER, THE LEARNING CURVE is primarily a useful estimating tool. While the concept provides a general model for the effect of practice on productivity, it does not account for variables such as employee motivation and resistance to change. Careful communications planning and management will help prepare employees for the rollout of a new system. The learning curve will help you design the training they need to use the system effectively. ■
Fred Borgianini, PMP, is a systems integration process specialist in the information technology organization of GTE. He is also a member of the Tampa Bay PMI Chapter.