The root of the cause

Managing the problems associated with the more manageable (and thereby accountable) business units.
global supply chain of major products requires a flexible, Risk is diversified across multiple “baskets” of suppliers. The
adaptable and consistent approach. benefits of local markets (such as cheaper labor and the prox-
One need look no further than the latest commercial aircraft imity to raw materials) can be exploited. The list goes on.
designed by Boeing to see just how far a company is will- But Boeing’s strategy also shows some of the inherent risks
ing to go to realize the numerous and enormous benefits of and unfortunate consequences associated with managing
a decentralized, extended global supply chain. Expertise and such a supply chain. Unfortunately, it is one thing to map out
specialization are focused directly on individual components. all the potential benefits of a diversified global supply chain,
Profit and loss responsibility is concentrated into smaller, yet quite another for the company and its managers to make it

38 Industrial Engineer
Solving supply chain problems proactively by chris eckert and brian hughes

out alive given the risks involved. encountered in such a diverse system can expect to rule their
While the new 787 Dreamliner by Boeing may be one of sectors. The key component to proactive problem solving is a
the most ambitious attempts at wringing benefits from an robust solution management system (SMS) built on a solid,
extended global supply chain, we can find other examples of adaptable root cause analysis program.
all sizes, shapes and flavors. Globalization is just another way The following is a set of simple steps and guidelines for
to say “global supply chain.” Like it or not, it is here to stay. establishing such a system. It explains how learning from fail-
Companies that learn to manage the risks of a global supply ures in the supply chain is the best way to minimize proactively
chain can expect to reap, at the very least, the reward of survival. the risk of future failures. These guidelines will help establish
But those that learn to manage proactively the problems or improve a solution management system. If you already have

February 2010 39
the root of the cause

a strong SMS that achieves the desired results, you may want Therefore, metrics need to be developed to provide real-time
to skim down to the more advanced steps at the end. There’s information showing, at a minimum, the following items:
something here for everyone.
• Number of units
Step 1. Map the supply chain • Estimated time to finish lot
This may seem to be an obvious first step. In fact, many compa- • Target time to finish lot
nies have an accurate understanding of their supply chain — at • Rejects and the reason for each reject
least at a high level. However, inside the larger process steps are • Downtime and the reason for downtime
a series of smaller steps. The key is to have each function map
out the part of the supply chain for which it is responsible. Again, keep it simple. Some of the best systems supply real-
time information to the operator that helps to focus attention
• Inputs: Raw materials (for example) on critical deviations. But not every company operates at such
• Where do they come from? a level. Do not develop a system that drowns the user in data.
• How do you ascertain their quality? And do not feel like you need state-of-the-art technology in
• What is the target delivery time? order to develop effective key performance indicators. Regard-
• Process steps: less of the sophistication of the technology used to capture and
• What happens at each step? report information, operations staff needs to see where break-
• What is the target cycle time for each step? downs occur.
• What other functions or suppliers interface at each step?
• Outputs: Step 4. Identify triggering criteria
• What does the process create? Getting accurate performance information from your process
• What is the throughput for the process? is important. But in order for that information to drive behav-
• What is the quality of the output? iors, you need to have simple, clearly defined thresholds
• How quickly is it being delivered to the next step? (triggers) that initiate a root cause analysis (RCA). Triggers
should be based on the information supplied by the key perfor-
The trick is to keep it simple, manageable and consistent. mance indicators determined above. Obviously, the supply
It should be easy for one manager to explain her process to chain is going to be most sensitive to problems at bottlenecks
another in a short amount of time. And be sure to involve because they determine throughput. So start with the bottle-
knowledgeable front-line employees in creating the process necks you identified in Step 2. Then move on to other steps,
map. They have specific process awareness that is important always keeping in mind throughput goals and their impact on
to capture. Securing their buy-in through involvement early the supply chain.
is important because they will play a crucial role in later steps. It is important to note that setting threshold levels is one
variable in a critical equation regarding your solution manage-
Step 2. Manage bottlenecks ment system. It is often difficult for managers to resist setting
Once you understand the process inputs, steps and outputs, extremely low thresholds because they want to be problem-
you must manage throughput. Without launching into a free. What is wrong with this approach? Low triggers create
discussion on the theory of constraints, it is important to iden- larger numbers of formal investigations. Managers chronically
tify the bottleneck steps in your process. These process steps overestimate the bandwidth of team members to investigate
are critical to your ability to meet production commitments. problems above and beyond their regular responsibilities,
Therefore, an effective solution management system must particularly in today’s climate of “do more, faster and better,
focus on eliminating problems affecting these bottlenecks. with less.”
But let’s face it — like all good things, a quality investigation
Step 3. Identify key performance indicators takes some time, both from the investigator as well as from the
Now that you understand your process steps and bottlenecks, team of required experts. A formal, consistent RCA methodol-
the next step is to develop or refine key performance indica- ogy helps ensure this time is used in the most effective way and
tors to determine whether you are meeting your production helps to determine where realistically to set thresholds.
targets. Problem visibility is required to solve the problem and Setting thresholds that are too low relative to the investiga-
is enhanced when viewed in light of production deviations. tive capacity of an organization will flood the system. If low

40 Industrial Engineer
thresholds are needed, additional investigative capacity must instance, if the product is a pharmaceutical product, appropri-
be provided. So be prepared to back off on the thresholds or to ate categories may be patient safety, worker safety, regulatory
invest in developing additional facilitators. At the end of the impact and product quality.
day, the value of achieving the production goals of the supply This list is not exhaustive, and it does not account for
chain should drive this decision. discrete impact values that may be important in any given
investigation. Frequency of recurrence needs to be docu-
Step 5: Implement a consistent mented. Risk assessments also should be conducted at this
RCA methodology point.
Most companies do RCA, but few do it well. Even fewer actu- Do not shortcut the process by skipping RCA Step 1.
ally have a robust, sustainable and effective RCA program in Making sure everyone involved understands the purpose of
place. In other words, few programs are integrated with their the investigation, as well as the impact on the supply chain, is
work processes and are reaping the potential risk reduction, critical to everything else that follows.
cost savings and return on investment. The first four steps RCA Step 2: Identify causes. An effective investigation is
above make good process sense. Operations people in the best not solely brainstorming possible causes. Brainstorming, à la
companies have these steps down to a science. But no process the fishbone technique, is a crude form of analysis that risks
is perfect; every system breaks down at some point. wasting time just when time is of the essence. The most effec-
Recurring failures show us opportunities for systemic tive way to identify causes is by starting with the problem
improvement. Failures present us with unique insight into identified in RCA Step 1. Then, using a logical process, the team
where we can eliminate risks in the supply chain and make deconstructs the problem to identify its underlying causes.
it more robust, but only when we have a comprehensive RCA At a high level, it works like this:
program in place. This does not have to be difficult or expen-
sive. But in order to achieve great results in today’s supply • An effect is the result of at least two or more causes. These
chain, you must do more than brainstorm around fishbone causes combine at a point in time resulting in the effect. This
categories or ask “why” five times. What is needed is a simple, implies that every time you ask “why” of an effect, you find
yet robust, RCA process upon which you can develop a proactive multiple causes, all of which are logically required in order
capability to reduce the risk of future supply chain problems. for the effect to exist.
RCA Step 1: Problem recognition and definition. • Every cause also is an effect, which means that as the inves-
Problem recognition is achieved by using the triggers devel- tigator continues to ask why, more causes are found. The
oped in Step 4 above. Every significant deviation from defined result is a diagram that looks like a tree on its side, with the
goals requires a formal RCA. If you’ve done a good job defining trunk on the left and the branches developing to the right.
triggers and you have balanced your investigative capacity at an • You can keep asking “why” until your diagram is large
appropriate level, problem recognition should be easy. Segments enough to represent the problem accurately.
of the supply chain should be assigned to various investigators
who watch for triggers to occur and then investigate. An important aspect of developing a logically sound cause-
and-effect diagram is to recognize two different types of causes.
Once an investigation has been triggered, the investigator,
with help from the team, needs to create a problem statement Some causes are catalysts; they trigger a change in condition.
by documenting specific information regarding the problem. They can be thought of as variables in the causal equation
At a minimum, this information needs to include the followingbecause their timing is often difficult to predict. For example,
items: a triggering cause of a broken bottle is “bottle dropped.” This
cause is momentary and transferable; it can happen any time to
• The problem title any person holding a bottle.
• The threshold triggered Other causes are more stable over time. These causes act
• When the problem occurred as constants in the causal equation and are found in the
• Where the problem occurred in the supply chain conditional environment. Continuing the example above,
• The impact on the supply chain other causes of a broken bottle are “bottle material” = “glass”
and “bottle elevation” = “five feet.” Both of these causes are
Impact is often relative to a specific supply chain link. required for the outcome of the event to be the effect of a
Impact categories often can be identified ahead of time. For broken bottle. Therefore, they should be represented in the

February 2010 41
the root of the cause

analysis as in Figure 1. inclusion of causes in the analysis. Sometimes the avail-

Both types of causes play a role in an event — the condi- able evidence is not very good. Other times, it is as solid as a
tional environment and initiators of change in condition. A rock. The goal of the investigative team is to uncover the best
good investigator, with the right training and a little practice, evidence. The primary reason for finding solid evidence is to
can achieve a high degree of proficiency at representing both ensure the investigative team has confidence in the accuracy
types of causes in an analysis. This is important for at least two of its representation as well as in the solutions it identifies.
reasons. First, any threshold-triggering problem is important. RCA Step 3: Identify solutions. Once the causes have
In such a case, it is essential to develop a thorough under- been identified, along with their logical relationships and
standing of the problem. Second, a thorough understanding supporting evidence, the team then examines the causes to
of a problem’s causes allows the maximum opportunity to find opportunities for solutions. Effective solutions control
identify effective solutions, which is the goal of any investiga- causes. Individually, a single solution reduces the risk that the
tion. Often, these solutions can be fast, cheap, easy and highly problem will recur. However, when multiple solutions control-
effective — always welcome qualities in today’s economic ling multiple causes are identified, the risk of recurrence drops
environment. even more substantially. And the risk drops the farthest when
causes from the conditional environment (the constants in
the causal equation discussed above) are controlled. These
broken bottle causes are consistent over a period of time. If you change the
conditional environment, you reduce the base-level risk that
Bottle affects everyone operating in that environment.
over Continues...
concrete RCA Step 4: Implement solutions. It is nothing less than
Bottle
a shame when a team does a good job investigating a problem
struck and identifying solutions, and then those solutions fall by the
concrete
wayside in favor of the newest crisis of the day. It is true that
Broken Bottle
bottle dropped Continues... many solutions never get implemented. This happens more
Bottle
often than members of a supply chain like to admit. Inherent
material= Continues... in a robust RCA program is the ability to implement solutions
glass
in a timely fashion. This includes ensuring that the solutions
implemented do not cause other problems in the supply chain
Bottle
elevation= Continues... and evaluating the effectiveness after an agreed-upon period
1.5 meters
of time.

Figure 1: In this example cause diagram, red causes trigger a

change in condition, and green causes represent the conditional
Step 6: Identify systemic risks
environment. An effective solution management system delivers more than
risk-reducing solutions for individual problems. It creates
a history of problem causes represented in a consistent
This logical determination of causes continues until the common format that can be mined for systemic opportunities.
investigative team has developed a thorough understand- A systemic problem is a nugget of elevated risk in the supply
ing of the problem. This means that the team accurately can chain. For instance, if one organization in the supply chain
explain what happened and identify an effective list of solu- has a high rate of turnover that is playing a role in multiple
tions that reduce the risk of recurrence. The investigation may problems, this high turnover rate represents a systemic risk
not find all the causes, nor does it need to. The goal of the to the entire supply chain, and it needs to be addressed. Other
investigation is to explain the problem to others and to reduce common systemic problems include forgoing maintenance on
the risk of recurrence through implementation of solutions critical assets in lieu of production, low employee morale, late
that break the causal chain. Once this has been achieved, the shipments of crucial raw materials, and so on.
investigation can be closed. A systemic issue is a common cause for a recurring series
Developing an accurate, logical representation of the causes of problems. Suspected systemic issues can be pinpointed in
is important, but there is another important step. The causes individual analyses during review by causal analysis experts.
need to be supported with evidence. Evidence supports the They can then be confirmed by examining a sample set of past

42 Industrial Engineer
analyses to find out if the suspected systemic issue has played Include feedback loops that allow for continually improv-
a recurring role in past problems. ing the process. This is a living process; continual evolution
Once systemic issues are identified, their causes can be should be engineered into it and encouraged over time. And
examined in the same manner as described above. Removing everyone needs to buy in for the process to work, starting with
these causes reduces the risk of future recurrence. Identifying the upper management in each link. Building cross-functional,
and mitigating systemic issues is the best way to prevent future cross-organizational work teams can help ensure consistency
supply chain problems proactively because, left unchecked, and keep everyone on the same page.
these systemic causes will show up over and over again in
future problems. Overriding principle
The process will work best when you, your suppliers and your
Step 7: Report/share results customers are all engaged as part of the problem-solving team.
If each organization in the supply chain were to master the Yes, this is scary and risky, but history has proven repeatedly
above steps, there is no doubt that it would reduce both prob- that it pays great dividends. After all, do you really believe your
lem frequency and impact. But the supply chain will benefit even customers put any stock in the “root cause” of “operator error”
more when a simple, effective reporting process is included. and the typically accompanying solution of “retrained operator”
Reporting memorializes the work of the investigative team. or “disciplined operator”?
When done well, it helps leverage the maximum benefit from When you bring your customer and suppliers together to
the efforts of the team. complete the RCA on the problem, you will find the following:
Too often, links in the supply chain remain compartmental-
ized. This is understandable. Teammates on one project may be • All parties own some part (causes) of the problem.
competitors on another. Plus, it is not easy to admit failures — • Each party discovers very real and important constraints
let alone thoroughly document and share them with the entire that other parties have to live within and deal with. This
team. Unfortunately, the entire supply chain suffers as a result. increased understanding leads to better collaboration and
Some of the most egregious manifestations of the detriments teamwork.
of compartmentalization can be found in hospitals. Errors • Suppliers appreciate the opportunity to be equal partners
often are driven underground for fear of litigation, which slows in the problem-solving process rather than recipients of
the spread of important learning opportunities, ultimately grenades tossed over the fence line. They develop more trust
increasing the risks to anyone requiring medical attention in in you and exert more energy to find truly effective solutions.
any hospital. You will begin to see less retraining and discipline and more
Learning from failures in the supply chain and sharing that substantive solutions.
learning through the organization is the best way proactively • Customers see that you are serious about improving the
to minimize the risk of future failures. Effective and timely quality of your goods or services, and they usually are more
reporting helps facilitate this in the short term. The best understanding and cooperative with you as you implement
opportunities should be singled out as case studies, presented your solutions. Who wouldn’t like a bit more of that these
at conferences and published in trade periodicals — all facili- days? d
tating long-term success.
Chris Eckert is president of Apollo Associated Services. He is a profes-
Step 8: Scale across supply chain sional engineer, a certified maintenance and reliability professional
Everything described so far is scalable. You can build a proac- and formerly worked as a reliability engineer with Dow Chemical and
tive solution management system at any link in the supply Rohm and Haas.
chain. Before rolling out a process like this across the entire
supply chain, though, it is best to ensure that it works on a Brian Hughes is vice president of Apollo Associated Services. He
reduced scale. The entire process can be piloted at an individ- has led significant incident investigations, including those related to
ual link in the chain. This allows for modifications to be made major explosions, chemical releases, consumer product contamination,
before any broader rollout. Flexibility will be required; today’s manufacturing defects and supply chain processes.
global supply chain involves different companies, countries,
governments and cultures. However, the steps defined here
are constants that should exist in some form at each link.

February 2010 43