Five Sigma

Process Improvement for the Less Compulsive

Return to Articles listing

What business person today has not heard of 6-Sigma? It has become the de facto standard in process improvement and enjoys the seal of approval from such notable firms as Motorola and General Electric. And it caps a long history of evolution in process improvement. We have come up from quality circles to process improvement to continuous process improvement to BPR to TQM and now to 6 Sigma. And 6-Sigma Black Belts can command premium salaries in the marketplace. One would almost get the impression that if you are at all serious about operational excellence, that 6-Sigma is a must.

There is, however, another point of view.

There are some features of 6-sigma that seem a little suspicious. It seems rather extreme for some businesses. Worrying about 3 or 4 errors per million may make sense if you are producing automobile parts for Honda, but not if you are selling computer security services to mid-size companies. It is not even clear what 3-sigma would look like! If I am manufacturing TV sets, the notion of “error” seems pretty straightforward. If the power supply fluctuates wildly, there is an error in the unit that can be easily remedied. But what is the definition of ‘error’ for a credit union processing auto loans? Aside from errors in transcribing information, the real errors are twofold: giving a loan to someone who does not pay it back, and refusing a loan to someone who would have paid it back. The first error would not be evident for months, and the second would not be trackable at all since the loan was not granted! If I cannot even count the errors, why would I worry about such an extreme demand for accuracy? And the worst error of all would be to focus on processing accuracy and ignoring the less accessible errors that really matter.

And when did process improvement become quasi-religious? Black belts sometimes act as if they had been anointed rather than trained. If you question the validity of their methods, you are more likely to get a knowing smile (“He hasn’t seen the light yet”) rather than a thorough explanation. While much of 6-Sigma is well articulated, some key assumptions are presented more like articles of faith rather than empirical features of operations.

For many businesses, the real strategic advantage is not in doing what they do better, but in choosing what to do in the first place. Six-sigma assumes you already have the right work processes in place, and the real business challenge is in squeezing out all sources of variation. As a systematic approach to achieving strategic goals, 6-Sigma has much to offer; if your strategic goals are ill-formed, 6-Sigma would be a dangerous distraction.

For all of its dominance in American business, 6-Sigma also has had its share of failures. Who has not heard of a company that invested heavily in a 6-Sigma program only to see it stumble and fall. While it may be powerful, it is apparently not ubiquitous. What are the circumstances where 6-Sigma would not be an appropriate intervention? And why not?

These are all legitimate questions, but it is too easy to toss out the baby with the bath water. A more thorough and careful assessment of 6-Sigma would distill the essential value without getting lost in the mountain of statistical tools and dogma.

For starters, let us distinguish between the tool kit and the craft. Practitioners of 6-Sigma often come with an impressive (intimidating?) array of methods and techniques. Statistical analysis of error. Pareto diagrams. Wishbone diagrams. Control limits. And the dreaded Design of Experiments! However powerful, they are not the essence of 6-Sigma, just the evidence of it.

Whatever the limitations of 6-Sigma, it is built around some key assumptions and beliefs that remain valid even if their implementation sometimes fails. Anyone considering adopting some version of process improvement would do well to include these basics even if they adapt the application or the tool kit.

The first assumption is the most obvious:

Process is a better way to think about work than jobs and departments.

Business people often underestimate the significance or the implications of this statement. The overwhelming history of business management in the United States has been a continuous re-application of hierarchical and modular views of work. The enterprise is carved up into functionally specialized departments which then divide activities into chunks that can be assigned to a particular position. Rewards and accountability focus on individual performance; presumably the work of many hands will roll-up to a total enterprise with the same logic that lower-level finances can be rolled-up to the financial report for the company overall.

The assumption is that people at the top have the greatest understanding of the work, and that people at the bottom would only need to “do their job” and it would all add up to a successful company. While this was the essential genius of Ford’s assembly line, that view of work has been eclipsed by rapidly changing technology, the need for innovation, the ascendance of customer service, the advent of the information age, and (of course) process improvement.

Perhaps the most frustrating implication of this belief is the futility of simple cost control attempts. Only a modular view of work would support a “10% across the board” reduction in staff or budget. A focus on process contradicts the frequent executive edict to simply “rein in the costs” when business gets tough. Cutting costs without consideration of the process implications is a sure fire method of hobbling key work activities and undermining the organization’s ability to retain customers and remain profitable. The only real savings are achieved through refining processes, which inevitably leads to lower cost per unit of value.

This first notion leads to the second key assumption reflected in 6-Sigma:

The senior support for process improvement goes beyond “sponsorship”. Senior executives have to think with process, not just manage process.

Lots of earlier process improvement efforts failed because senior executives treated it as a resource for middle-level managers or operational teams while they continued to view the organization and govern it with their usual concepts and procedures. As process improvement pushed for integration or for changing legacy systems, the executives became the greatest obstacle to process improvement. Assignments and rewards continued to be allocated to functional heads, so those driving process change found themselves politically adrift and pitted against powerful adversaries (department heads) with little real authority to carry out their process charters.

Six-Sigma responded to this shortcoming by defining the role of executive champions and by requiring that 6-Sigma projects be linked to substantial quantifiable dollar improvements.

But before the 6-Sigma strategy will prove productive, executives have to translate strategic objectives into their requisite process implications rather than assigning them to departments. The assignment to department heads is in terms of the processes on which they are expected to cooperate. It is the processes of the company that are considered able to produce value, not departments.

Most organizations have found the friction between departments as a significant upper limit to overall performance. Legal won’t work with Sales in generating quick turnaround on bids. Engineering doesn’t support Customer Service in addressing technical questions. Manufacturing won’t work with Engineering to produce more workable product designs. IT is notoriously slow in providing customized applications despite pressing needs. While every department has a compelling rationale for their behavior, the bottom line is that the overall enterprise suffers. Process improvement pulls for attention to work flows across departmental boundaries, which compromises the authority of the department head and enhances the authority of the process owner.

The net result is that adopting a 6-Sigma approach will create more political challenges than it does technical challenges. The tool kit of 6-Sigma can be learned; the shift in mind set in the senior ranks requires a conversion experience.

The third assumption reflected in 6-Sigma is a difficult one for most managers to swallow:

The worker knows more about the work than the manager.

The hierarchical structure of traditional business has always vested authority in the “boss”. The supervisor or manager is supposedly the expert on doing the work. But the attention to process reminds us that the people with their hands on the levers actually know more. With today’s rapidly changing technologies, managers can become out-of-date in a very few years. This does not mean that employees end up “in charge”; it only means that their first-hand experience with work processes deserves a significant voice in process assessment and re-design.

The challenge to the manager is to shift their role from “command-and-control” to “facilitator”. Rather than telling people what to do, they are required to set up a certain type of conversation, one which is anchored in customer requirements, oriented to process flows, and responsible to quantitative measures and statistical analysis rather than managerial presumption. And the manager has to make sure the conversation extends across the full range of the process, not just that part of it within his or her departmental boundaries.

The fourth assumption in 6-Sigma is fundamental to the attention to process at all:

Errors are overwhelmingly a reflection of poor process design rather than individual error.

The almost dogmatic insistence on individual accountability that is so prevalent in today’s business runs squarely up against a key belief in process improvement. Although most process errors will have a human face on them (if only the person who reports the error), the source of the error is overwhelmingly in the process and not in the person. This simple fact is easy to miss if you have been steeped in the American traditions of hierarchical authority and individual initiative. We have such a powerful belief in the power of the individual and their proper accountability for their actions, we often miss the tangled web of actions that make up a process. We are like the drunk looking for his lost car keys under the lamp post where there is plenty of light, rather than down the block where he actually dropped them.

The last key assumption of 6-Sigma is not hard to accept, just hard to put into practice:

The 6-Sigma methodology of process improvement (DMAIC) is more powerful than common sense, managerial judgment, or informal problem solving.

The overall improvement strategy of 6-Sigma (Define, Measure, Analyze, Improve, Control) is a superior alternative to simply rolling up your sleeves and “dealing with the issues”. It provides a rigorous and disciplined sequence that makes process improvement data-based and continuous. It also imposes new skill requirements on those who would pursue serious advances in process performance. Rather than assuming a room full of well-motivated and smart people are sufficient, the DMAIC methodology sets the bar a little higher.

We should acknowledge that DMAIC is a new set of terms for a well-established sequence of thought. It is a minor variation on the PDCA (Plan-Do-Check-Act) sequence that grew out of Juran’s seminal work. And the fundamental lesson is still the same: we do not really know what we do until it is measured and tracked!

Even this cursory summary of the essential tenets of 6-Sigma reveals some useful clues about why it might stumble. For many organizations, the prevailing culture simply may not allow it. If error is always treated as a reflection of individual shortcoming, we will not find the process problems behind the individual stumbling. If your first response to an error is “Who screwed up?” rather than “What went wrong?”, process improvement (in any flavor) is not likely to take hold.

Such a culture will also create a skewed sense of the frequency and location of errors, since it discourages the revelation of mistakes. In contrast, a 6-Sigma culture would view error as a corporate resource to be exploited.

Cultures that reinforce the authority of the manager will not only attract managers who enjoy wielding such power, it will also suppress the insight and subtlety known only to the front line worker. Quality Circles (one of the first steps toward quality) can work only when employees feel it is safe to discuss errors publicly. Without that safety, process improvement will be like fertile seeds cast on hard, dry ground.

Managers who have clawed their way to the top of their department will not welcome the intrusion of process owners into their domain. They have grown to expect they will receive marching orders and an associated budget with which they can “run their own shop”. The demands created by process improvement for cooperation between departments and for committing resources based on process needs will seem a serious threat to the department head looking for autonomy and unquestioned authority.

Finally, some 6-Sigma efforts will fail simply because the senior executives fail to understand their role. At one of my client companies, quality concerns were relegated to Friday afternoon discussions, since senior managers insisted on the rest of the week to “do their work”. As soon as they separated ‘quality’ from ‘their work’, the effort was doomed.

6-Sigma started out primarily in manufacturing. The physical and contained nature of the work flow made a process orientation easy to cultivate. The high volume of work and the simple definitions of error made quantitative analysis a powerful tool. And 6-Sigma has argued that process improvement can be extended to other processes in the company as well. Clearly it could be easily extended to any activity that involved “assembly” of information rather than materials. Order taking, warehouse operation, or financial activities should all be amenable to 6-Sigma techniques.

But are there activities where 6-Sigma would be inappropriate?

Consider product development. Certainly there is a data flow and a financial layer that could be enhanced with 6-Sigma. But these are supporting activities; the fundamental value-add of product development is the cultivation of innovation. Redundancy would be an error in production processes; if innovation is the goal, redundancy might be a valuable stimulus to new thought. Experimentation and even serendipity require a looser process.

Or what about consultative selling? I could certainly apply 6-Sigma to the order process, but the fundamental value of consultative selling is creating a relationship characterized by rapport and trust. ‘Defects per million’ is not very meaningful when I have 12 clients over a period of a few years. The focus is not on the accuracy or speed of any particular transaction, but on its contribution to the relationship which creates the promise of long-term business.

And how about strategic planning? While components of strategic planning need to be data-based, the most important outcome is the full engagement of all participants. Even if the plan itself is only 90% of the possible quality, the better implementation achieved through employee engagement is worth many times more than improving the quality of the plan. The process by which people are brought to the table might be more important than the final product. Once again, the emphasis of 6-Sigma could be a dangerous distraction.

For a final example, take sales forecasting. At first blush this would seem a perfect candidate for 6-Sigma. Information from last year’s sales are combined with data on shifting demographics, size of customer base, or competing customer choices to produce a forecast of next year’s sales. Based on that information, the company knows how much raw material to order, where to expand or contract their sales presence, how much expansion they can legitimately support, or where to build storage facilities. Where someone is applying judgment, the more we could extract their expertise into the process, the better it would be.

Unfortunately a sales forecasting process has to survive a political process. The forecasted sales translate into financial growth, which would determine the BUY / HOLD / SELL recommendation of financial analysts. Senior executives would have to defend the projected growth rate to the Board. In one of my client companies, the sales forecaster was told “to find more growth”. Despite the sophistication of his forecasting model, the executives simply insisted on double-digit growth to satisfy Wall Street and ensure their own bonuses. In such a distorted political environment, the real challenge is not to refine the process, but to confront the executives who want to tweak the numbers for all manner of reasons.

There are some characteristic features of these potential exemptions to the application of 6-Sigma:

• The processes are not designed for high volume.
• The definition of an ‘error’ is not simple; some errors are not visible until much farther down the process.
• Errors are not modular and easily corrected; some errors contaminate the entire process and cannot be remedied (consider the impact of a flawed sales forecast!)
• The process outcomes are more social and emotional than physical or informational.
  We have no reason to expect the same actions would have the same result over and over.
• There is a significant component of judgement based on experience rather than expertise.
• The processes are not contained; they are vulnerable to a range of unknown and uncontrollable external influences.

This does not mean the principles of process improvement are irrelevant, only that the heavy-duty statistical tools of 6-Sigma are less likely to add any real value. The participants of these processes would do well to characterize their efforts as a flow, even though key steps may be a black box that cannot be opened or extrapolated in any meaningful way.

Whatever the shortcomings or strengths of 6-Sigma, the concern for (1) process as the proper unit of business and for (2) the ongoing improvement of processes as the shortest path to profitability is perfectly sound. The variance is in how to apply the basic principles of 6-Sigma as a function of the level of process focus in your organization. So let’s start with some basic distinctions in the level of process focus and what form of process improvement makes sense in that circumstance.

Level 0: Traditional, hierarchical structure; employees expected to “do your job”

In this type of organization there is probably little use of the terminology of process. When it occurs at all, process capture is likely to be very procedural in focus and contained within a particular department. It is likely to be more prescriptive than investigatory, that is, it is the manager’s attempt to specify behavior required rather than learn from the actual behavior expressed. Employees may fear that it is an attempt to place blame or to constrain their behavior even more. For example, a new “travel reimbursement process” is likely to be imposed from above as a control mechanism.

The best approach in such an organization is to recognize that process improvement will be first and foremost a challenge to the culture. Select a supportive manager with a compelling need to improve process flow. Provide ample support and expertise. Use the effort as a pilot experiment to introduce new ideas into the company culture. Make sure senior support is genuine, visible, and durable. Be wary of using process characterization as a precursor to reducing headcount; it may sour the organization to any further process improvement efforts.

It will be critical to capture the procedures for process improvement as well as the procedures for the process in question! The focus may be on better inventory management, but the way the group goes about the investigation will be even more novel. Make the steps of process improvement visible through constant reference, explicit capture, and training.

Level 1: Processes defined within departments.

In this organization the concept of work process may be well known, but it is still exercised within the clean hierarchy of a department. Expertise and authority is probably still vested with the manager, with the employees providing information and suggestions. Managers have not had to re-think their role much.

The real challenge is linking processes across departmental boundaries. Find a process with a clear link to an important strategic objective; you will need a compelling argument for threatening turf and departmental authority. Make sure the two managers involved are on board and willing to speak with one voice to their respective staffs. Make sure they understand that they cannot improve the process by improving “their piece” of it; they have to demand and exemplify cooperation across department lines.

If necessary, start with a less threatening examination of the handoff between the departments. One of my clients refused to let me use the “P-word”. They had lost over $1M a few years earlier in a botched attempt to train everyone in TQM. My contact told me flat out that using the P-word in the proposal would run afoul of the CEO. I asked if we could use the H-word; after a puzzled look I explained that the H-word was ‘hand-off’. Apparently the CEO liked that word, so we brought departments together so they could explore the hand-off of information or materials from one department to another. Just so everyone understood each other well, I asked the initiating department to explain the steps it took to generate a particular report. To complement their contribution, I asked the second department to explain a little more of how they used the report once they received it. While everyone found the explanation of the entire sequence most helpful, we had managed to avoid the P-word. (To any experienced outsider, it was obvious that we had just defined the entire process; the hand-off was just the starting point to work further back up stream and then further down stream in the process.)

In less testy organizations, you can leverage the concepts of process more directly. The difficulty will be enlisting cooperation across the departmental boundary. Often the most resistance comes from the department heads. Staff are often farther along since they have forged informal cooperative agreements just to get their jobs done. The shift in the authority of the department head needs to be highlighted as a significant culture change. As the organization pushes for greater cooperation at that level, be prepared for noticeable turnover. Some department heads will opt to become process owners; others will simply opt out.

To avoid the political threat to department heads, there will be some pressure to define a process as a new department. For example, someone might propose a Customer Complaint Department. Supposedly they could handle the “whole process” from complaint to resolution. In operation, of course, this process will require technical information from Engineering, credit adjustments from Finance, or product returns from Warehousing. Ideally processes are defined from the perspective of the customer, not for the convenience or comfort of department heads.

This level of process focus will probably show uneven development across the organization. Manufacturing probably embraced process concepts some time ago; Sales or Legal or Product Development may resist application because “they’re different”. While different kinds of activities may, indeed, require different styles, eventually the entire company has to embrace some agreement on the value of being process focused.

Level 2: Key processes defined across the whole enterprise.

At this level, process concepts and methodology are well established. Managers have adjusted to the need to be more collaborative with their peers, and even to work under shared accountability for process performance. A process improvement effort should be fairly standard and not be worth particular fanfare. In fact, it is important that it be treated as “business as usual” and not receive any special treatment. The shift is to continuous process improvement rather than one-time efforts such as business process re-engineering.

The next step is to install the role of the process owner as someone who has authority and control of resources. If process owners are always subservient to department heads, processes will eventually suffer as departmental agendas steal away key staff, focus, or resources. Raising the authority of process owners often creates considerable turmoil at the senior level. The challenge is political in nature rather than technical. The resolve of the CXO’s in driving the change will be sorely tested. The benefits, however, are significant. Senior leadership will finally have some control over the enterprise, rather than always having to settle for managing the pieces.

Level 3: Senior leadership is a mix of department heads and process owners.

In this most advanced stage, process improvement is well established. Process becomes the unit of thought in building the business. The rights and roles of the process owner is no longer a matter of major debate (although it will probably never be completely smooth). They are accepted as deserving a place at the senior table, and not relegated to operational management.

The next step will be applying the principles of process improvement to governance rather than just operations. The processes of the governance council (it is too big now to call it a ‘team’) should be captured and made subject to the same review as that used for lower level activities. Strategic planning needs to flow directly into process requirements rather than functional objectives. A set of metrics is needed to track process performance against requirements. Problems are immediately traced back to the relevant processes rather than to individuals or departments.

The second challenge will be to turn process improvement back onto process improvement. The process of improving processes can be improved. This second order learning should lead to a richer set of process tools. When process improvement feels unwieldy or cumbersome, the outcome should be a revised and refined strategy for process improvement. Perhaps some work activities are best captured as a story board rather than as a flow chart. Perhaps for some processes it is more important to capture criteria for outcomes than precise procedural steps. The goal is to avoid making process improvement into a religion to be slavishly applied in some rigid sequence. It is a tool kit, and the craft and the product are more important than the specific tool. Any creative or even heretical use of the tool which yields the desired result is well within the spirit of process improvement. And that is more important than being within 6-Sigma of the mean.