How many times have we felt the same way, bumping along as always and wondering whether there is a better way to do things? Each time we are faced with a familiar problem, we swear to ourselves that we will look deeper into the situation and really solve it once and for all. But then something else comes up, causing us to push the question to the back burner until it surfaces again.

Events, Events, Events…

We live in the world of events:, “The stock market dropped 15 points today… A five-car pileup occurred on interstate 95…Our copy machine broke down at 3 o’clock…Our first quarter earnings were down by 20%…Our latest product launch was 10 weeks late…” and on it goes. When events (like a car breakdown) have a direct impact on our lives, we have to react as quickly as possible to them. But there is no long-term leverage for creating change in an organization if we only stay at the level of events (see “Levels of Understanding: ‘Fire-fighting’ at Multiple Levels,” June/July 1993).

For example, managers at A-to-Z Corp., a semiconductor company, have been puzzling over a series of events that occurred in their most recent quarter. They posted record sales for the quarter, with the majority of the sales force meeting or exceeding sales quotas. All products scheduled for release were launched, with additional products ready for early release in the next quarter. At the same time, however, profits actually declined for the first time in company history, as overhead costs as a percent of sales reached an all-time high.

Interrelated Patterns of Behavior

The frenetic pace at A-to-Z made it easy for its employees to get caught up in the daily demands of the semiconductor business. Until their profits declined, A-to-Z’s managers had no idea that there might be underlying financial problems. To address the issue of falling profits, they decided to collect data about their past performance, looking back over a period of time to identify important patterns of behavior. What, for example, was the pattern of product launches over the last two years? Or the number of new products in the pipeline? The number of product engineers? The average experience level of engineers?

The data they found was as follows: Sales revenue had risen every quarter for the past 10 years, but profit growth had been falling for the last several quarters and had actually declined in the most recent quarter. Because the company’s past success was based on new products, there was a continued commitment to launching many new products each year. Each quarter they added new sales people to meet more aggressive sales targets.

An initial plot of these events over a period of several years is shown in “A-to-Z’s Performance OverTime.” Just plotting the data, however, provided little insight about why the trends have occurred.

Data Analysis and Theory Building

There are many tools and methods available for analyzing such time series data as the A-to-Z managers collected. The quality improvement arena, for example, offers run charts, scatter diagrams, and statistical process control methods for analyzing trends, interrelationships, and system capability. Various regression or trend analysis tools are also available for identifying correlations between variables. But there are limitations to the use of these tools.

A-TO-Z’S PERFORMANCE OVER TIME

Sales revenue at A-to-Z has risen every quarter for the past 10 years, but profit growth has been falling for the last several quarters. Meanwhile, new product launches and the size of the sales staff have been increasing every year.

One obvious limitation is that regression or any other data analysis tool is useless without data. However, there is often a paucity of data available for analyzing a new problem — and therein lies the dilemma. If we are only using data analysis tools, we run the risk of just focusing on those variables for which we have data. On the other hand, it is unrealistic to try to track data for everything in advance. Data analysis tools and methods are most useful when they are used as part of a theory-building process.

Drawing behavior over time (BOT) graphs (also called “reference modes”) can help break the data availability dilemma by building causal theories before we gather the necessary data. The BOTs can be used to connect past observed behavior with future behavior in a way that offers insight into the causal structures underlying the case. Developing such causal theories reduces the risk of becoming straight-jacketed by the limitations of the data that is readily available. In short, BOTs guide the use of data, but are not data-bound.

Building a Theory

The A-to-Z managers began working with their initial behavior over time charts by putting together a cross-functional team to try to understand what was happening. This team decided to look at a time horizon of five years.

To begin to understand why profits were falling even as revenues were growing and new product introductions were running smoothly, the A-to-Z team hypothesized about the relationship between total number of new products and the unit cost of carrying products. Although the number of products in their catalogs had been growing steadily, they wondered whether the cost of carrying the products was growing at an even faster rate. One person inferred that the number of new products with revenues of less than $10K was probably increasing and that the average selling price was decreasing.

This possibility would help explain how they could have record unit sales and dollar volume and still have falling profits. Another person suggested that increasing revenue pressures might be putting pressure on new product development to keep pumping out even more new products. These pressures might cause people to work on creating products that were easier to develop and launch, rather than on more innovative and potentially more profitable ones. This emerging causal theory is shown in “Pressure on New Product Development.”

Guidelines for Using BOTs

As the A-to-Z team members continue to work with and build confidence in their causal theory, they can begin to gather the appropriate data to see whether it supports what they have theorized using the BOTs and the accompanying causal loop diagram. Through an iterative process of going back and forth between theory-building and data analysis, they can build a better understanding of what is happening.

When you begin using BOTs and causal loop diagrams to build causal theories of specific issues, some general suggestions can help guide the process:

1. PRESSURE ON NEW PRODUCT DEVELOPMENT

1. 1. Select Time Horizon. Identify the desired time horizon for the problem at hand. The length of time will provide a guide for determining which variables to select and study further. Having a time horizon of two years, for example, will have different critical variables than those associated with a time horizon of 20 years.
2. 2. Define the Problem Dynamically. Draw behavior over time charts of key variables. These charts can serve as reference points throughout the theory-building process, helping to define the problem, focus the conceptualization, and validate the emerging theory.
3. 3. Conduct Thought Experiments. Conduct thought experiments by hypothesizing about the time behavior of different variables and inferring the behavior of other related variables. Do “what-if” experiments of possible future scenarios and draw out the implications of those events on other variables.
4. 4. Build Causal Theories. Use causal loop diagrams to build causal theories that draw out the interrelated behavior of variables over time.
5. 5. Validate with Data. Use data analysis tools to help validate the BOTs and causal relationships.

If we don’t want to be like Edward the Bear — forever bumping our heads down the stairs — we need to be able to step out of the day-to-day stream of events and see the larger context in which we operate. Drawing behavior over time charts and a corresponding set of causal loop diagrams can not only show us what happened, but can also help us build a better understanding of why something happened.

The post Behavior Over Time Diagrams: Seeing Dynamic Interrelationships appeared first on The Systems Thinker.

In the face of day-to-day pressures, groups often leap to solutions after only a modest amount of brainstorming. A systemic approach, however, provides a structured problem-solving process for digging deeper into our most vexing problems.

To get a sense for how systems thinking can be used for problem identification, problem solving, and solution testing, we have outlined a six-step process. To use this process on a problem in your workplace, try the worksheet on page 9.

1. Tell the Story

The starting point for a systems thinking analysis is to get your head above water enough to start thinking about the problem instead of just acting on it. An effective way to do this is to gather together all of the important players in the situation and have each one describe the problem from his or her point of view.

At Bijou Bottling Company, the problem was usually a customer complaint: “Where were the 40 cases of 2-litre Baseball tie-in product that were ordered last week?!” Somehow Bijou would get the goods there on time, whatever it took—including air shipping heavy soda in glass bottles at enormous costs. But this crisis management led to a culture where people built their careers on coming in at the 11th hour and turning around a customer complaint.

2. Draw “Behavior Over Time” Graphs

In the storytelling stage, most of the energy is focused on the pressures of the current moment. When we move to “Behavior Over Time” (BOT) graphs, however, we begin to connect the present to the past and move from seeing events to recognizing patterns over time.

Draw only one variable per graph on a Post-it note so it can be easily moved around in the steps that follow. The time frame should span from past up to the present—but it can also include future projections (see “Bijou Over Time”).

3. Create a Focusing Statement

At this point, you want to create a statement that will help channel energy during the rest of the process. This statement may involve a picture of what people want, or a question about why certain problems are occurring. At Bijou, for example, the focusing statement was: “We’re pretty good at solving each problem as it arises. But why are these problems recurring with greater frequency and intensity? What is causing them?”

BIJOU OVER TIME

At Bijou, crisis management efforts had increased over time, while the effectiveness of the production/distribution system had decreased.

4. Identify the Structure

You now want to describe the systemic structures that are creating the behavior patterns you identified. The systems archetypes are an easy way to begin building a theory of why and how things are happening (see “Systems Archetypes at a Glance,” V22N6, August 2011).

Begin by reviewing the story, graphs, and focusing statement to see if they follow the storyline of an archetype. If so, draw the loop diagram for that archetype, place the Post-its of the variables in the diagram, and move them around on a flip chart until you have a diagram that seems to capture what is going on.

The group at Bijou decided that their problem matched the “Shifting the Burden” storyline, in which a problem is “solved” by applying a short-term solution that takes attention away from more fundamental improvements. They identified a balancing loop that described how customer problems were solved with heroic “11th-Hour” efforts (the symptomatic solution) at the expense of improvement and redesign of the production/distribution system (the fundamental solution). As people “learned” over time that heroism is rewarded, their willingness and ability to address system-wide problems decreased (see “Shifting the Burden to Heroism”).

SHIFTING THE BURDEN TO HEROISM

At Bijou, customer problems were solved with heroic “11th-Hour” efforts (B1) rather than with improvements in the production/distribution system (B2). Over time, people at Bijou “learned” that heroism is rewarded, which reduced their willingness and ability to address system-wide problems and increased the company’s dependence on heroic efforts (R3). One negative side-effect of Bijou’s “heroism” attitude was that customers were taking problem situations and escalating them to crises in order to get the company’s attention (B4).

5. Going Deeper into the Issues

Once you have a reasonably good theory of what is happening, it is time to take a deeper look at the underlying issues in order to move from understanding to action. There are four areas you should clarify:

• Purpose of the System. Ask yourself, “In the larger context, what do we really want here?”
• Mental Models. Begin the exploration of mental models by adding “thought bubbles” to those links in the diagram that represent choices being made (see “Mental Models and Systems Thinking: Going Deeper into Systemic Issues,” V23N5, June/July 2012).
• The Larger System. Add links and loops to enrich the story and connect the relationships to the larger system.
• Personal Role. Acknowledge and clarify your own role in the situation.

For example, when the people at Bijou looked at the larger system, they wondered what role their customers played in the system. They theorized that customers were taking problem situations and escalating them into crises in order to get the company’s attention (B4).

6. Plan an Intervention

When planning an intervention, use your knowledge of the system to design a solution that will structurally change it to produce the results you want. This might take the form of adding a new link or loop that will produce desirable behavior, breaking a link or loop that produces undesirable behavior, or a combination of the two. The most powerful interventions often involve changing the thinking of the people involved in the system.

At Bijou, the key to change was realizing that the problems were largely self-inflicted. They realized that they had to make progress on production/distribution system improvements while still doing enough fire-fighting to keep things afloat. In the longer term, they would need to change the reward systems that promoted heroic behavior. They also recognized the need to sustain the improvement efforts even when the pressure came off—otherwise the problems would be back again soon.

Part of a Cycle

Even as systems thinkers, it is easy to fall back into a linear process. But learning is a cycle—not a once-through process with a beginning and an end. Once you have designed and tested an intervention, it is time to shift into the active side of the learning cycle. This process includes taking action, seeing the results, and then coming back to examine the outcomes from a systemic perspective.

Michael Goodman is an internationally recognized speaker, author, and practitioner in the fields of systems thinking, organizational learning and change, and leadership.

Richard Karash is a founding trustee of the Society for Organizational Learning, a founding member of the SoL Coaching Community of Practice, and a co-creator of “Coaching from a Systems Perspective.”

Editorial support for this article was provided by Colleen Lannon.

The post Six Steps to Thinking Systemically appeared first on The Systems Thinker.

Next, do a quick “clean-up” of the variables by getting rid of “positive” or “negative” qualifiers (e.g. “good,” “bad,” etc.) and stripping away action words (verbs).

TIP: If you must choose a variable that is either positive or negative, it is preferable to select the positive sense—for example, it is better to use “growth” rather than “decline” because it is clearer what increasing or decreasing growth would look like.

Placeholder Terms: Peeling the Onion

In the beginning stages of loop-building, it is often easiest to lump multiple concepts together in a single “placeholder” term while you sketch out the rest of the story. For example, “Employee Investment” represents a broad category of investments, including salary, training, and morale-boosting activities.

At this point, therefore, you may want to ask, “Of the terms that I lumped together, are there key issues that should be pulled out separately?” You may feel that a decrease in your training budget, for example, has a significant effect on your company’s service and sales—so you may decide “Training” should be included as a separate variable. The process of going over the loop again and again to clarify the variables is similar to peeling an onion, revealing deeper layers of issues. How deep you go depends on the specific issue and on what level of understanding you want to gain.

Iterative Process

After you have worked with the diagram for a while, you can begin to fine-tune the variable names to clarify the picture. For example, you may ask yourself if there is a clearer way to describe the variable “Employee Investment.” Suppose employee investment in your company depends upon the size of the human resources budget—it would therefore be clearer if the term “Employee Investment” were changed to “HR Budget.”

Expect that your loops will go through many drafts as you continually clarify the story. There are some additional guidelines that may help you select appropriate variable names:

• Use nouns. Avoid verbs, action phrases, or terms that suggest a direction of change, since the “action” in a CLD is conveyed in the arrows. For example, “Decreasing Sales” will cause confusion when you read through the diagram and ask what happens when “Decreasing Sales” increases or decreases. “Sales” is a better choice.
• Variables should be quantities that can vary over time—things that can rise or fall, grow or decline. “Sales Staff Turnover,” for example, is preferable to “Sales Staff Perceptions” (perceptions can change, but they usually do not increase or decrease).
• Is time used in any of the variables? Time itself should generally not be included as a causal agent. When something changes over time, it generally does not change because of the passage of time.
• In drawing CLDs, it is often useful to make a distinction between actual and perceived states. You may find that integrating “actual” or “perceived” into your variable names will help you to clarify your diagrams.

It takes many iterations to create a good diagram, especially if it contains several reinforcing and balancing loops. It is often helpful to show your loops to others to gain different perspectives and enrich your understanding of the dynamics. Another person can help clarify a diagram by pointing out links that are confusing, or ones that may have been missed. Remember, you are not mapping “truth,” but your explicit understanding of how a system operates.

Kellie Wardman is vice president of the Greater Manchester Family YMCA. She also serves as adjunct faculty in creative writing at Southern New Hampshire University. Kellie was publications director of Pegasus Communications. She holds an MFA in creative writing from Emerson College. Go to her blog at http://kelliewardman.com.

The post Selecting Variable Names for Causal Loop Diagrams appeared first on The Systems Thinker.

The “Fixes That Fail” Storyline

In this structure, a problem symptom gets bad enough that it captures our attention; for example, a slump in sales. We implement a quick fix (a slick marketing promotion) that makes the symptom go away (sales improve). However, that action triggers unintended consequences (diverts attention from aging product line) that make the original symptom reappear after some delay—often worse than before.

Some people know this dynamic from mismanaging their finances. Whenever they run short of cash, they use their credit cards to “solve” this shortfall. Unfortunately, the additional debt increases their monthly credit-card payments, causing them to run short of cash the next month. They again “fix” the problem by using their credit card to cover an even greater shortfall (because more dollars are going to pay the finance charges on the debt). Many juggle their debt among several credit cards by paying one card off with checks written on another. But with each round, the debt burden grows heavier and heavier, which may be why we currently have the highest consumer debt levels in history and record personal bankruptcies—all in a booming economy! This is the basic storyline of the “Fixes That Fail” archetype Let’s take a closer look at how and why this systemic structure behaves the way it does.

Of Heroes and Scapegoats

Many managers report that their organizations experience certain problems over and over again. Most seem to accept these challenges as a fact of life. Only a few see the cause as “hard-wired” into their businesses. However, from a systemic perspective, whenever patterns of behavior recur over time, they must be driven by structures that are designed into the way the system operates—intentionally or not. To better understand why we would create such structures, we need to take a closer look at the behavior of this archetype (see “The Hero-Scapegoat Cycle”).

THE HERO-SCAPEGOAT CYCLE

When a problem symptom becomes a crisis, we look for a hero to drive it back to acceptable levels using quick fixes. By the time the unintended consequences of those fixes cause the problem symptom to reach crisis level again, we’ve promoted the hero. We therefore scapegoat the new manager for failing to keep the problem under control.

Organizations usually have target levels against which they monitor performance; for example, inventory levels. If a problem symptom exceeds its desired level, such as excess inventory, we may notice this trend but not act on it right away, because we’re focused on other, more dire crises. When the symptom eventually reaches crisis proportions, we then shift our attention to that problem. At this point, because the situation has become so dire, we often look for someone who can “save the day” (e.g., slash inventory levels). Sure enough, we find a person who can drive the symptom down to the desired level in a hurry and then reward her with a promotion.

In the meantime, the delayed consequences of the hero’s actions (lack of product availability due to low inventories) begin to have an impact, and the problem symptom returns (higher inventory levels). When it again reaches the crisis level, we blame the person who is currently overseeing that function for failing to do his job, fire him, and look for our next hero. However, in this archetype, it may well be that the first hero is the person who put the current crisis in motion and that the scapegoat is the person who set the stage for a more lasting solution to take hold. But, because of delays in the system, these realities are often obscured.

Win Today, Lose Tomorrow

So, why do so many organizations fall into the “Fixes That Fail” trap? Why can’t people recognize the vicious cycle that keeps repeating the same patterns of events? One of the reasons is that the delays in the system mask the true nature of the cause-and-effect relationship. The narrow time frames that often drive decision-making in organizations also compound the problem.

For instance, our results are more likely to deteriorate over time if the delay for the unintended consequences to affect the system is long than if the delay were shorter. This is because, without the feedback supplied by the unintended consequences, the “improvements” actually appear to make things better in the short term (see “Fixes That Fail over Time”). And yet, when we view the situation from a longer time horizon, we find that today’s “desired” levels are far higher than yesterday’s “crisis” levels. From a longer perspective, we see that those short-term successes are part of a series of steps toward longterm failure. This pattern shows how companies can go bankrupt even as individuals are continually rewarded for doing a great job.

FIXES THAT FAIL OVER TTME

Over the short term, we applaud the progress we are making. And yet, when we view the situation from a longer time horizon, we find that the current “desired” levels are far higher than yesterday’s “crisis” levels used to be. Thus, those short-term successes are actually part of a series of steps toward long-term failure.

What Alarm Bells?

Another problem associated with this archetype can occur even when we do make changes so that quick fixes are no longer needed. Now, this may sound like a good thing, but it all depends on how we do it. Unfortunately, many organizations solve the problem by adapting to the poorer performance level, which then becomes the new norm (or desired level).

For example, we may have had a desired first-run capability of 95 percent or better from our production line (that is, 95 percent of our motorcycles run the first time off the assembly line), but we often find ourselves operating at a crisis level of only 90 percent. Because our plans are based on the higher level, our ability to provide predictable performance drops.

In order to improve predictability, we lower our desired level to one we know we can achieve (90 percent), with plans to eventually bring our capability back up to 95 percent. The danger of such a move is that once we have factored the poorer performance into operating plans, it becomes less visible as an issue that needs attention. In other words, what once caused alarm bells to ring no longer rings any bells, because we have in effect disconnected them. Although we no longer reach the crisis level or require frequent fixes, we have embedded the poorer performance in our system, and we no longer notice it.

In this situation, we have fixed our problem by getting caught in a different archetypal structure called “Drifting Goals.” We end up “fixing” things by changing our criteria of what constitutes a crisis.

Finding Fixes That Last

Of course, the answer is not that we should never apply quick fixes. There are many circumstances for which we absolutely have to implement short-term solutions. The danger lies in failing to recognize that all quick fixes are merely stopgap measures that buy us time to get to the root causes of those problem symptoms.

One of the most important points to address about this archetype is the relationship between the delay for the unintended consequences to show up and the timing of organizational performance assessments. If you suspect that you may be caught in a “Fixes That Fail” dynamic, look for a repeating pattern of quick fixes, determine how often these fixes occur, and compare that to the frequency with which you typically review performance. If the review time horizon is about the same as or shorter than the time between fixes, then try lengthening the time frame so that it’s at least three or four times the delay period. This will help provide a more accurate picture of the actual “progress” being made.

Daniel H. Kim, Ph. D., is publisher of THE SYSTEMS THINKER and a member of the governing council of the Society for Organizational Learning.

The post Fixes that Fail: Why Faster is Slower appeared first on The Systems Thinker.