Into the Design Factory
Imagine that own a business with a large bank account that pays no interest. You would be unhappy. Now imagine that this bank account loses half its capital value every year. You would be outraged. Yet this is exactly the situation that most companies find themselves in as they try to manage their product development process. They have high levels of investment in partially completed designs, but these investments earn them nothing until the products are introduced. These investments are at risk, because a competitor's product introduction can make a partially completed project worthless overnight. In effect, they have a large non-interest-bearing asset that loses its capital value very rapidly. There is however one key difference. Most companies have no clue how big this asset is, or what sort of return they make on it.
This chapter introduces the concept of the Design Factory, the foundation upon which we will build when we introduce more powerful analytical tools later in the book. We will look at the goals of the Design Factory, its output, its inventory, and will examine its similarities and differences from the manufacturing factory. I have chosen the term Design Factory for our product development process because it gives us a useful context within which to view this process. By approaching the problem as a manufacturing problem, we can exploit certain tools and concepts that have proved useful for repetitive processes like manufacturing.
There is a good reason for viewing design through the eyes of a manufacturing manager. In general, more careful management thought and more insightful review of experience have occurred in the area of manufacturing than in the area of product development. This is true in part because we have better accounting systems in manufacturing, which makes it easier to assess performance. It is also true because we spend more money in this area, causing it to receive a higher level of management attention and scrutiny. It may also be true because manufacturing is an inherently repetitive process and therefore more likely to produce learning than a nonrepetitive process, like product development. Whatever the causes, the net effect is that we have scrutinized manufacturing more carefully than product development. This careful study has sharpened our thinking and produced some useful tools that can usefully be applied to the development process.
OUR GOALS ARE ECONOMIC
We must begin by asking, "What is the objective of the Design Factory?" We cannot optimize the Design Factory unless we are clear about what we are trying to optimize.
The answer is quite simple. The Design Factory exists for one purpose -- the same purpose as the manufacturing factory -- to make a profit. I would argue that it has no important objective that cannot be quantified in economic terms. Furthermore, I would assert that it is only by attempting such quantification that we can objectively weigh the consequences of particular actions. We are only philosophers until we begin to use numbers.
This view is not as starkly mercenary as it appears. All profit derives from a single source: the ability to sell things or services for more than they cost to make or acquire. Our sales price is a measure of the value that people attach to our output. If we fail to create sufficient value for customers, then they will spend their money on something else. In this sense sales price is the customer's message that we have done something useful. Our cost, on the other hand, is a measure of how well we use time and physical resources. A manufacturer who is efficient at converting material, labor, and energy into products is rewarded with profits. When resources are squandered, profits drop. In this sense, profit is simply a measure of how efficient we are at converting time and resources into things people value. When we do important things for people, without wasting time and resources, we are rewarded with profit.
We will develop this measure of profit more completely in Chapter 2, where we will show how measures like time, expense, cost, and performance can be converted to a common denominator of profitability. For now, just remember that our objective is profits, not products.
PRODUCTS VS. DESIGNS
How does the Design Factory plan to produce these profits? Here is where we find a fundamental difference from manufacturing. Producing a design is a different problem from producing a product. As Marv Patterson has pointed out in his book, Accelerating Innovation, the purpose of a design process is to generate information. It is the difference between making food and making recipes. The manufacturing factory creates food for people to eat. The Design Factory creates recipes. Behind this simple statement are some interesting implications that we shall examine in more detail in Chapter 4 when we cover information theory. For now, let us make a few basic observations.
The recipes we produce in the Design Factory only achieve our objective when they are converted to our valued goal, which is profit. Thus, the only measure of the value of design is its economic value. Our designs are not important because they are new, or because they are beautiful, or because they are innovative; they are only important if they make money. Our designs will only make money if we create recipes to turn material, labor, and overhead into valuable functionality better than our competitors do.
The goodness of our recipe is important, but it is perishable. The ability of a design to make money is greatly affected by time. Our recipe is constantly measured against the recipes offered by other companies. The profit of the recipe is determined by its value, which takes the form of sell price, and its cost. Both of these factors vary with time. For example, the original Sony Walkman sold for around $150. Since it cost less than $150 to make, it was profitable. The same design today would make no money because the market price for this functionality is less than $30. Market prices have fallen because today's designs have different recipes. They have substituted plastic parts for metal ones and can be sold profitably at today's low prices. Thus, we need good recipes, and we need them before our competition gets them.
When we begin a design we make investments in creating recipes. However, a recipe does not generate profit until it is completed. During the time that the recipe is incomplete we are holding an investment that is earning no money. We call this investment design-in-process inventory (DIP).
This is the Design Factory's equivalent of the work-in-process inventory (WlP) that we have in the manufacturing factory. Like the level of WIP, the level of DIP is a sign of the health of our process. When we bring our designs to market quickly, we can begin earning money quickly, which is good. When we are slow, we will have large levels of DIP which earn no money for us. Thus, the design process has inventory just as the factory does.
This DIP inventory is actually more important than inventory in the factory, because it is much larger and more expensive to hold. Manufacturing plants turn their inventory with incredibly fast cycle times. Modern plants can turn their inventory more than fifty times a year. In contrast, it is the rare development process that achieves more than one turn per year. Thus, DIP can be a much larger number than WlP in most companies. As the data in the Figure 1.1 show, DIP can easily be ten times bigger than WlP in many situations.
Furthermore, holding costs are much higher for DIP than for WlP. A traditional rule of thumb for the cost of holding WlP is about 25 percent of the inventory value per year. This takes into account the cost of storage, the risk of obsolescence, the cost of capital employed, and so on. The holding costs for DIP are much higher because the risk of obsolescence is high. In certain categories, a product that arrives to market two years late will have zero value. Most manufacturing inventory does not drop to zero value nearly as quickly.
You would think that the high levels of DIP and the high cost of holding DIP would be incentives for management to measure and manage it. Sadly, it is ignored, due to a critical weakness of our accounting systems. For example, look in any annual report. Where is DIP? There is no evidence that it exists. But why?
The invisibility of DIP is actually a cruel trick played on us by Generally Accepted Accounting Principles (GAAP). Long ago, accountants agreed that R&D expenses should be recognized at the time the money is spent. They maintained that the value of a partially completed design could not be assessed with enough accuracy to treat it as a financial asset. They asserted that companies would inflate their balance sheets and overstate their profits by claiming that money spent on R&D was the same as deposits in a bank.
These arguments are sound, but the unfortunate consequence of such sound accounting practices is that DIP is invisible on our financial statements. Neither shareholders nor managers have any clue whether they are managing this resource well. Because we never show it on our balance sheet, we do not think of DIP as an asset to be managed, and we do not manage it.
Thus, an incomplete design is like an invisible non-interest-bearing checking account. You have invested the time and effort but you are not yet getting a return. Even worse, it is like a non-interest-bearing account that loses its capital very quickly.
The key value in recognizing the existence of DIP is that recognizing it empowers us to do something about it. If we don't know DIP exists, we cannot manage it. Later in this book you will see how development process design choices can minimize the magnitude and cost of this inventory.
RISING COST OF CHANGE
We have seen that the Design Factory has profit objectives and inventory just as the manufacturing factory does. There are, however, areas where the Design Factory can be quite different from the manufacturing factory. One such area is the way the cost of change behaves during the process.
The cost of making changes during product development rises exponentially throughout the design process. This is quite different from a manufacturing process, where value-added rises more linearly through the process. In a design process, the cost of making changes early is exceptionally low, whereas the cost of late changes is very high. This exaggeration of economics will have important implications for how we need to manage the design process. Exponentially rising costs make it critically important to drive changes into the upstream portion of the design process, where such changes are hundreds of times cheaper to make. Figure 1.2 contrasts the cost of change in design and manufacturing processes.
Because of this rising cost of change the problem of managing changes becomes a very important one in the Design Factory. Late changes are very expensive. We shall see later in this book how this changes our approach to generating information and managing risk.
The irony is that while it is very easy to talk about driving all changes as far upstream in the design process as possible, it is much harder to make this happen. The key problem is that information becomes available later in the Design Factory than it does in the manufacturing factory. To see this we must make a subtle distinction between requirements information and information about execution.
In a manufacturing process, our product requirement is 100 percent complete at the beginning of the process. The requirements will almost never change during the process itself. At each manufacturing step, we try to get feedback about the effectiveness of our execution. If we have failed to meet an intermediate requirement, we take some sort of corrective action. For example, if we are machining a shaft we know the required dimensions the instant we start machining. When we perform individual operations on it we verify they have been done correctly, and fix them if they haven't.
Let us contrast this with what happens when we do product design. We think we have the requirements defined, but they always change in some way during the design process. This means that, unlike the manufacturing process, we receive some of our requirements after work has begun. We also get feedback from the customer about the suitability of our design solution after we make the design choices. Often this feedback comes very late, such as when we perform beta tests at the end of the design process. For example, we may design our shaft with a certain thickness. We may discover that the shaft breaks in field testing due to unexpectedly high loads. We now have to make changes very late in the design process. Thus, we are plagued in the design process with late arrival of information. Figure 1.3 shows the difference between the Design Factory and the manufacturing factory in the timing and arrival rates of information.
Interestingly this late information arrival interacts negatively with the rising cost of delay because a lot of information arrives when it is expensive to react to it. As we look at this problem more carefully later in the book we will discover that we do not have to passively accept this late arrival of information. Instead we can actively manage the timing of information arrival and thereby have an important impact on our economics.
There is another important difference between the Design Factory and the manufacturing factory. The manufacturing factory usually does things more than one time in a row. In contrast, the Design Factory primarily engages in one-time processes. In the Design Factory we will never do the same thing, the same way, twice in a row. The design process produces information, or "recipes," and there is no value in creating the same "recipe" twice. For a manufacturing factory, which produces products, there is value in producing the same product twice. This means that in the design process we can only add value when we do something differently. If we change nothing, then we add no value. In contrast, in manufacturing if we change nothing we can still make money. You could say that while we make no money by reinventing the wheel, we can make money by manufacturing the same wheel many times in a row.
This constant pressure to do something new and different has important implications for how we have to manage the process. It means that we will see much more variability in the design process than we see in repetitive manufacturing processes. This variability is an indicator that we are doing something new in the design process, and thus that we are adding value.
Unfortunately, many managers schooled in the techniques of variability reduction in their manufacturing processes have not carefully pondered whether such variability reduction is appropriate in their design processes as well. As we shall show in Chapter 4, risk, and the variability associated with it, are inherent and desirable characteristics of design processes. They are at the heart of the design process's capacity to generate information. The path to eliminating variability in design processes can easily lead to a process with no value-added.
This difference in the role of variability has important implications for the type of tools the Design Factory inherits from the manufacturing process. The manufacturing process, which is repetitive, benefits from variability reduction, so it has developed tools to reduce variability. The Design Factory needs variability, so it needs tools that allow us to coexist with variability. If we blindly use manufacturing factory tools in the Design Factory, we may do more damage than good. For example, in manufacturing processes we create metrics, we set standards, we measure deviations, and we take corrective action. We shall see later how the same approach can be poorly suited for the world of product design.
Fortunately, the unsuitability of manufacturing tools does not leave us toolless. Our manufacturing tools are well-suited for dealing with predictable, deterministic processes, but in development we have variable, one-time processes. The good news is that powerful tools have been developed for managing variable, one-time processes. We will describe these specialized technical tools in Chapters 3, 4, and 5.
There is one final important difference between our Design Factory and the manufacturing factory. This difference arises from the one-time nature of the design process, which we just discussed. A manufacturing factory normally does things more than once. Whenever we engage in repetitive tasks we have some implicit standard for how long the work should take. In effect, the finish line is defined for a manufacturing process. We have either finished making the part or we haven't. For example, we machine a piece of metal and measure it to assess if we are done or not. If we have achieved the desired measurement we stop.
In the very different world of engineering our work perversely expands to fill the time available. Some managers blame this problem on the engineers. In reality, it is inherent to the engineering task itself because such tasks require solving a problem that has not been solved before. When we try to solve a problem for the very first time, we consider various options, some likely, some far-fetched. We usually rule out the unlikely options and proceed to check the likely solutions. The classic model of design is explained in Nobel laureate Herbert Simon's famous work The Sciences of the Artificial. Simon's observation is that we do not continue searching forever for a perfect solution but rather we stop searching when we find a satisfactory one. His design world is bounded by reaching a satisfactory solution.
In a certain sense, this takes a strange form in the real world of engineering. Here, we rarely stop when we achieve a satisfactory solution, because this will simply result in a satisfactory product. Instead, we stop when we run out of time. You can see this for yourself by looking carefully at when engineering tasks are scheduled to be complete and when they are actually completed. In theory, if an engineering task was over when an engineer found a satisfactory solution, then some engineering work should be completed early and some engineering work should be done late. In practice, we see something very different: engineering activities are completed either on time or late. They are never completed early. When we do engineering, we stop when we run out of time. If we find a satisfactory solution early, then we use the extra time to find a better one, because there are always things we can do to improve performance or lower costs. We have little incentive to deliver our work early, because the next activity will probably not be ready to start early anyway. It is always more attractive to polish the design than it is to have it gather dust waiting for the next activity. Actual task completion times look very different from theoretical ones, as we can see in Figure 1.4.
This property of expanding work has important implications, as we shall see in Chapter 11. It means that the schedule will inexorably unravel on us unless we use a process that compensates for this expansion. Fortunately, there are ways to do this.
We have seen that there are many similarities between the Design Factory and the manufacturing factory. Both are primarily interested in making money. Both use input resources to produce output, although the output in one case is products and in the other case is information. Both factories have inventories, although for the most part our design-in-process inventory is hidden from financial visibility because of our accounting practices.
At the same time there are some key differences. In the Design Factory the cost of change rises dramatically during the process. This will affect the way that we think about change. In the Design Factory we start the process with less of the information that we need to complete it. This means that information is arriving in the middle of the process, which has very interesting implications for our process design.
Our Design Factory processes are much less repetitive than manufacturing, and have higher variability. Moreover, this variability is central to their success. We shall see later in this book how this variability is our friend and not our enemy. This fact drives us to seek tools designed to allow us to live with variability, rather than tools designed to destroy it.
We also observed that the Design Factory engages in inherently expandable tasks. The finish line is made of rubber and can easily move away from us. This will have important implications for how we structure our control systems, as we shall see later on.
We now have a general concept of the Design Factory and some of its characteristics. However, we must penetrate a bit deeper to understand how we can manage it. To do so we need to equip ourselves with some basic tools. These are covered in Part Two, Thinking Tools. In Chapter 2 we will look at economic modeling tools that give us quantitative insight into project and process economics. We will then look at the field of queueing theory in Chapter 3. Queueing theory allows us to understand process design and capacity management issues in a quantified way. Chapter 4 will introduce information theory, a powerful tool for understanding how a design process should produce information. Finally, in Chapter 5, we will look at systems and feedback theory to understand their implications for the design of complex systems. The concepts in Part Two of the book are a critical underpinning for the action tools we will describe in Part Three.
Copyright © 1997 by by Donald G. Reinertsen
Managing the Design Factory
The first book to put the principles of World Class Manufacturing to work in the development process, Managing the Design Factory combines the powerful analytical tools of queuing, information, and system theories with the proven ideas of organization design and risk management. The result: a methodical approach to consistently hit the "sweet spot" of quality, cost, and time in developing any product. Reinertsen illustrates these concepts with concrete examples drawn from his work with many leading companies across different industries.
Fresh and thought-provoking, the book challenges many of the conventional approaches to product development. "There are no best practices," Reinertsen writes, "the idea of best practices is a seductive but dangerous trap." Unlike other books that promote rules and rituals based on benchmarking "best practices," this book focuses on practical tools that account for varied situations. He breaks new ground with a disciplined, quantitative approach for making decisions on critical issues: When should we use a sequential or concurrent process? Centralized or decentralized control? Functional or team organizations?
Full of practical techniques, concrete examples, and solid general principles, this is a real toolkit for product developers. Moreover, it is written with the clarity, precision, and humor that are Reinertsen's trademarks. He promises to challenge the thinking of anyone involved in product development.