Optimizing Human Performance: an Introduction to the HPT Method
“I tell you, sir, the only safeguard of order and discipline in the modern world is a standardized worker with interchangeable parts. That would solve the entire problem of management.”
Jean Giraudoux, The Madwoman of Chaillot, 1945.
Ah, the French! They have a knack for getting to the heart of the matter. But until they actually develop that “standardized worker”, we’ll have to manage with the workers we have. They are anything but standardized, yet we have to fit them together without gaps or overlaps to produce an efficient, smoothly running organization. It’s a bit like working a jigsaw puzzle in reverse: instead of finding out where pieces go by how well they fit, we know where all of the pieces are supposed to go, but we have to do a little work to get them to mesh perfectly.
Not so long ago, the problem of getting the right fit was seen as a matter of selecting qualified candidates, training them, and then turning them loose to “do their thing”. This approach worked reasonably well when companies were prized for their stability and longevity. The best companies were the ones that made the same product, in the same way, year after year. Employees would be hired into entry-level positions and promoted bit by bit over a period of twenty years to leadership roles. People defined their careers as a progression within the company, expecting their pension (and a gold watch) as rewards for years of loyal service.
Today, the most prized companies are the ones that can anticipate and respond to changes in the marketplace. The best companies are flexible, reinventing themselves to meet new needs or employ new technologies. They survive by improving on their own products before competitors do it for them. In this environment, getting the pieces of the performance puzzle to fit is a continual task, part of the process of adapting to change. Instead of longevity, today’s goal is making your productivity and return on investment (ROI) greater than at any competing company. This is difficult to do because business processes are very complex, and because the criterion for success is constantly being raised by one’s competitors.
Faced with such a difficult task, only a systematic approach will succeed. This introduction describes a method that has been proven in actual practice, proven both in its feasibility and in its results. It describes the elements of a performance system that is regulated by feedback from both internal and external sources.
The pioneers of performance engineering, such as Frederick W. Taylor (1912) and Frank & Lillian Gilbreath (1917), perfected work study methods that put the behavior of workers under the microscope. More recently, Gilbert (1978), Mager & Pipe (1970), Rummler & Brache (1990), and Panza (1989) have come up with much easier methods that are effective without being as intrusive. The new methods, collectively known as Human Performance Technology (HPT), focus on the goals of the organizational structure and the accomplishments produced by workers’ behavior, not the behavior itself. As a practical matter, it is much easier to assess the results of a process than to continually observe the actual performance of the entire process.
Except for cottage industries and artists’ studios, most businesses require many workers and processes functioning harmoniously to operate the business. When we analyze performance systems, we treat the component processes as individual cottages coupled together. The accomplishment output from one cottage becomes the raw material input to another (Figure 1). The quantity and quality of the product is regulated by feedback from the customer to the supplier.
Figure 1: The Simplest Performance System
A feedback loop provides information upstream about conditions downstream. This information provides the means to maintain quality and adjust the rate of production. By adding additional feedback loops to the system, feedback about market needs can be used to modify the product. If any of these feedback loops is missing or dysfunctional, that part of the system is an “open loop”. It is free-running, unregulated by the needs of the rest of the organization.
Performance systems can be viewed as several layers. The top layer, the enterprise view, is the macrosystem that regulates the operation of the enterprise as a whole. With few exceptions, all enterprises have the same major components at this macrosystem level (Figure 2):
Figure 2: The Enterprise View
Our suppliers provide materials and services to us, and we provide them with specifications, orders, and feedback to tell them what they need to change. Similarly, our customers order our products and services, and give us feedback on how well we meet their needs. We have to pay attention to other inputs as well: what our competitors are offering, what regulatory agencies require of us, and how much money our bankers will let us invest in the business.
The Core System
The next layer down, the organization view, shows the operation and regulation of functional groups within the organization, representing the internal departments within “Our Organization” as suppliers and customers. For example, here is a company that
makes widgets (Figure 3):
Figure 3: Organization View of Widgets Inc.
In this diagram, the direction of the arrows indicates the supplier-to-customer relationship. While it may look a lot like product flow, it’s not the same thing. For example, in this particular company, the Sales Department has only customers, no suppliers, because it has been tasked to manage customer relationships, not the entire sales process. As a result, it receives feedback from customers and from shipping, but it has no control over them. That’s the sort of problem that the organization view can reveal. In a real company, things are more complicated because there are more departments, and multiple supplier-customer relationships between functions. For our purposes, this model is sufficiently complex.
Metrics and standards
Figure 3 shows the structure of the system, but that isn’t enough to tell us how well it operates. To do that, we must define performance metrics and establish criteria for each supplier’s output. For example, Manufacturing supplies finished goods to Inventory. An appropriate metric for Manufacturing’s output would be the elapsed time between Inventory placing an order, and Manufacturing delivering finished goods.
For each metric, an acceptable range, a standard of performance, must be defined. In this example, the criterion might be set at less than two weeks for some products, but less than three days for critical spare parts. Similar metrics and standards for time, cost, rejects, errors, customer satisfaction, etc. are established for each supplier’s output, and define the relevant dimensions of the supplier’s performance. If we focus on quantitative results, we can only manage what we measure, so our goal should be to measure and track the accomplishments of each functional group continuously.
Just making the measurements doesn’t provide control. The information must be delivered, that is, fed back, to upstream entities in usable form, and appropriate action must be taken. In our example (Figure 4), Sales surveys customers and generates sales forecasts by product for Manufacturing. Manufacturing places orders for raw materials with Purchasing. Purchasing informs suppliers of anticipated demand, and so on. The supplier-customer connections between Marketing, Engineering and Manufacturing form the feedback loop with the marketplace to modify product specifications.
Figure 4: Feedback Loops in Widgets Inc.
How many feedback loops are required? As a general rule, we’ll need one loop for every type of variation that we want the organization to respond to automatically, without intervention by senior management.
As with customer-supplier relationships, feedback loops that are missing, misdirected, or provide ambiguous information can cause problems. Too many feedback loops can be as bad as too few. If they are contradictory, they cancel each other out. The most frequent problem areas are the linkages between Sales and Credit, Sales and Customer Service, Purchasing and Customer Service, Manufacturing and Sales, and Manufacturing and Engineering.
For example, an in-house travel agency was supposed to provide good service to company travelers, and at the same time reduce travel expenses by routing them on airlines with negotiated rates. But “good service” to its customers meant finding the most convenient flights, regardless of airline. The travel agents resolved this bind by giving the effective complainers exactly what they wanted, so that “customer satisfaction” remained high, and putting everybody else on the “Brand X” airline. This solution carries the seeds of its own destruction; in time, everybody will be motivated to complain. Identifying such impasses and resolving them properly can easily justify the expense of this analysis.
Going into detail
Some performance engineers, having completed the organizational analysis, will stop at this point and go home. That would leave us with a performance system that functions down to the department level. Operating with such a system, all we can do is reward or punish department heads, relying on them to figure out how to obtain the desired performance. There are companies with self-directed work groups that do just that, but they are a definite minority. Most line managers will insist on a more precise system, especially when they have detailed goals that are not being met.
Performance microsystems extend the same concepts down another level or two, so that every worker’s performance is measured and rewarded individually. This requires a shift in focus, from who is doing what, to how it is being done. It takes time to define and measure individual accomplishments, but it makes troubleshooting much easier. Being able to pinpoint problems and fix them as soon as they are detectable pays big dividends in efficiency, flexibility, and morale.
To illustrate a process view, here is a small portion of the process diagram for manufacturing widgets (Table 1). The functional units are the rows, and the columns show the sequence of steps. This portion represents the turning and finishing of widget spindles in Manufacturing:
Table 1: Process Diagram for Widget Spindle Manufacturing
|Stock room||Cut rod stock|
|Tool room||Supply blades, oil||Supply bits, gauges, oil||Supply grit wheels, compound||Supply lube|
|Turning||Turn bearings, grooves|
|Grinding & Finishing||Grind & polish|
The stock room supplies 0.375″ by 6″ steel rods to the lathe operator, who turns two retaining grooves and two bearing surfaces on each one. The spindles are delivered to the Grinding & Finishing Department, where the bearings are polished and the ends ground flat. The tool room provides carbide tool bits, cutting oil, measuring gauges, and cleaning rags.
The lathe operator’s performance is measured against production standards for the number and quality of spindles produced per shift. The grinding operators check each lathe operator’s output for quantity and quality just as the lathe operator provides feedback to the stock room to regulate its output of spindle blanks.
When we come down to the task (or job) level, we evaluate the performance of each step. Given the current manufacturing method, is there potential for improving the lathe operators’ performance? There are several methods for answering this question. The easiest one is to compare each operator’s performance with other operators doing the same job, and compute the productivity difference between the best operator, called the exemplary performer, and every other operator.
Performance Improvement Potential
The sum of these differences is the Performance Improvement Potential (PIP). If we plot the average number of spindles produced per shift and the average number of rejects, we can compare operators (Figure 5):
Figure 5: Spindle Production
Operator G is clearly an exemplary performer. He has the greatest production, and the fewest rejects. The difference between his performance and the other operators is (Figure 6):
Figure 6: Performance Improvement Potential
If all the operators performed at full potential, output would be increased by 16%, and rejects would be reduced by 26%. To improve performance, we’ll carefully study what Operator G does, and compare him to the less productive operators. If he knows something they don’t, we’ll train the other operators to do likewise. If they know how, but don’t do it, we’ll try to remove the barriers to full performance.
Value Added Performance Measurement
If we don’t have a valid local performance standard, we can use performance benchmarks. Looking at other manufacturers using the same equipment, we learn that a good operator can average about $154.65 in value per hour of operation. We also know that a widget spindle costs $15.83 when we have obtained bids from outside suppliers. If we subtract the costs of the raw materials, we can compute the value added by the process of transforming spindle blanks into spindles. If the raw materials cost $1.13 per spindle, the value added by turning is: $14.70. When we multiply this value by their output and subtract the cost of their rejects, here is how our operators measure up (Figure 7):
Figure 7: Relative Value of Production
Operators G and B are better than the benchmark, but Operators A and F are substantially worse. These results demonstrate that the equipment can perform at the expected level, but some of the operators can’t, or won’t. From a management perspective, the operators “below the line” are not delivering the productivity that was expected when the machinery was purchased, so the return on investment (ROI) suffers.
If you compare the value method to the PIP method, you’ll notice that a single value metric encompasses both output and reject measure of performance, and that “value added” benchmarks are more transportable (between specific tasks) than exemplary performers. On the other hand, you don’t need cost information external to the situation to employ the PIP method.
Once we have documented a performance problem, how do we end the analysis and go about solving it? That’s our next task but first, there is one more analytical step to be taken. We have to determine why the problem persists.
Solving Performance Problems
Once we have identified a performance problem, we have to resist the temptation to solve it at the policy level, by writing a memo and circulating it to “all concerned”. There are a lot of causes for poor performance, and the best cure is the one that addresses the root of the problem, as well as the contributing factors. The answers to a few well-chosen questions will reveal the appropriate starting point.
Is there a simpler way?
The first question to ask is: “Can we make the process easier to accomplish?” If the process is difficult, either physically or mentally, it is a candidate for simplification. Because simplification usually reduces the time required to complete the task, it is the first option to consider.
Is there a skill deficit?
If workers lack the skills that the process requires, it could be that they have forgotten some seldom-used aspects, or it could be that they never learned them. The solution is to provide highly focused training, practice, or just more feedback, depending on the situation.
Does good performance produce positive consequences?
Consider the consequences that workers experience when they perform the task correctly. Are they punished, rewarded, or ignored? If reporting a safety hazard is considered to be “making trouble” for management, reporting them will happen less and less often because most workers will avoid being labeled as “troublemakers”.
In our Widget spindle example, Operator G was not only very experienced, but he had acquired a hearing loss at high frequencies. He liked to turn spindles at a faster rotational speed that the other operators, who found the high-frequency noise especially objectionable. So, part of that performance gap was closed by increasing the flow of cutting oil to reduce cutting noise and insisting that everybody wear high-quality ear protectors. The moral of this story is: don’t assume that the difference between “exemplary performers” and mediocre performers is just skill and knowledge. Sometimes exemplary performers are just better adapted to their environments.
If there is no consequence one way or the other, why should workers do the job in anything but the easiest possible way? Good performance, and only good performance, should produce positive consequences.
Are other tasks or methods more rewarding?
Few jobs these days involve only one task. When some of their tasks are unpleasant or boring, workers will find reasons to leave them undone, especially when there are no consequences. This also applies to the methods they choose. Classic examples are leaving safety shields off equipment, and “forgetting” to wear safety glasses. The solution is to remove any environmental obstacles to good performance, and to make the rewards for doing it right greater than the rewards for doing something else or doing it in some other way.
This multilevel, targeted approach to performance analysis and improvement avoids the futility of the shotgun approach – such as the infamous memo from “On High” that tells one and all: “The floggings will continue until morale improves.” It also avoids much of the delay and expense of “re-engineering” every process in the entire company. It does require substantial data collection and analysis, but it doesn’t try to fix things that aren’t broken. Notice also that in our problem taxonomy, only a few problem types actually call for training. The rest require changes to the work environment, or to the reinforcing consequences. With most real-world problems, more than one problem category will be relevant, so the solution will require a combination of performance interventions to produce the optimum solution.
Using the HPT method, optimizing performance is a process that starts at the top of the organization and works its way down to individual tasks. This approach has been learned through bitter experience, because every client tries to steer us immediately to individual workers, without pausing to consider problems that have been built into the working environment. It’s only natural; wanting to solve their problems in the quickest and easiest way, clients focus on the most variable and adaptable element in the performance equation: the human performer.
But human behavior has immutable laws of its own, and like Newtonian physics, they operate everywhere, all the time. Permanent changes in performance only occur when the principles governing behavior are applied correctly and consistently. Wishing it were otherwise, as so many managers do, will never make it so.
Gilbert, T. F. Human Competence. New York: McGraw-Hill, 1978.
Gilbreth, F. B., and Gilbreth, L. M. Applied Motion Study. New York: Sturgis & Walton, 1917.
Mager, R. & Pipe, P. Analyzing Performance Problems. Belmont CA: Fearon, 1970.
Panza, C. M. Picture This … Your Function, Your Company. Convent Station NJ: CMP Associates, 1989.
Rummler, G. A. & Brache, A. P. Improving Performance – How to Manage the White Space on the Organization Chart. 2nd Ed. San Francisco: Jossey-Bass, 1995.
Taylor, F. W. The Principles of Scientific Management. New York: Harper’s, 1912
Shneider, E. W. (2019) Optimizing Human Performance: an Introduction to the HPT Method. Being Better Matters 11 November. Available at https://www.beingbettermatters.net/optimizing-human-performance/ (accessed: date of your access)
Ed is the proprietor of Peacham Pedagogics, a consulting firm founded in 1984 to specialize in research and development for interactive instruction. In recent years, it has expanded into design, development, and evaluation of performan ce systems as well.Over the years we've seen a wide range of contexts: banks, supermarkets, Air Force avionics, Army tracked vehicles, publishing, audio-visual learning laboratories, foreign language acquisition, municipal operations, music schools, etc.