More Recommendations on Part Numbering

Three years ago, a previous post made the case for the key approach to nomenclature, as opposed to the obsolete “smart” numbering systems. In the key approach, the only job of a part number is to be a unique item identifier, through which all relevant information can be retrieved from a database. But you still need to think what items you want to have unique IDs for.

Continue reading

The Lowdown on Lean Accounting

Lufthansa’s  asked about it in the TPS Principles and Practice discussion group on LinkedIn. The following is the responses I posted, with a few links and illustrations added. The topics covered are as follows:

0.1. Scope of Lean Accounting

When applying Lean to any support activity, we must consider separately the impact on effectiveness and efficiency. Ram’s comment on “eliminating redundant activity, streamlining work processes,…” addresses the efficiency of the accounting department’s internal operations.

Instead, what Lean Accounting is primarily about is the effectiveness of the accounting function in supporting the business. It is about the nature of the services provided, and it only concerns management accounting — that is, the numbers that are intended to provide management with information on the effects of its actions.

Accounting also has to provide reports to external stakeholders in formats and with contents that are mandated. It is a game of using the rules to make the company appear simultaneously poor/struggling to tax authorities, and rich/successful to investors. But these numbers do not tell a plant manager how he or she is doing.

0.2. Background on Management Accounting in the US

In this as in many other areas, the practices that are standard today are relics of an era in which the work of manufacturing was more manual than it is today and data processing was manual. 100 years ago, for example, unit costs for manufactured products were tallied on paper spreadsheets and comprised of materials, labor, and overhead, with overhead being allocated to products in proportion to the labor consumed in making them.

The information technology of 1900

Information technology ca. 1900

Today, every high schooler carries more computing power than the J.P. Morgan bank had in 1900, and a steel mill uses barely 1/10 of the labor it required just 50 years ago to support the same volume of production.  It spends about ten times more on overhead than on direct labor, and allocating this  gigantic overhead to products based on a tiny labor content has been recognized as unnecessarily simplistic in the management accounting literature for decades, yet it dies hard in factories.

For at least 150 years, the most advanced work in management accounting has been done in the US. As Alfred Chandler tells the story, in the 19th century, accountants defined business-specific metrics — like “cost per ton-mile” for railroads — that are useful for companies focused on only one business. Today’s airlines, for example, use “revenue passenger miles.” Later, diversified companies needed generic tools to compare divisions engaged in different businesses, leading 100 years ago to a system of ratios like Return on Assets or Return on Investment, developed at DuPont and still in use today.

Dupont system of business ratios in Wikipedia

Alfred P. Sloan implemented the DuPont model at GM in the 1920s, and it is credited in part with the company’s success in competing with Ford, where nothing of the kind was done until the post-World War II “Whiz Kids” era. The vocabulary of the DuPont model now permeates business,  and is often used incorrectly. Just about Return On Investment (ROI), I recently saw the following two statements:

  • “XYZ Consulting helped clients achieve more than $317M in ROI.” An ROI is a ratio, not a sum of money. It is expressed as a percentage, not in currency units.
  • “We have an ROI of 2 years.”  Again, the ROI is a ratio of yearly values. Its inverse, in years, is the Payback Period.

American management’s love affair with these tools culminated in the 1950s, when executives from the financial side took over control of large corporations like GM and Ford, and managed them like car drivers who keep their eyes on the instrument panel rather than the road. In the 1953 movie Executive Suite, the R&D manager and the chief accountant of a furniture manufacturers compete for the job of CEO in a board meeting. In the movie, the R&D manager wins over the directors with a passionate speech about products, quality, and pride in workmanship; in real life, on the other hand, the accountants won.

By the 1980s, the problems this caused were obvious, and critics raised their voices. Harvard’s Robert Kaplan explained how management accounting had lost its relevance, and Eli Goldratt denounced accounting as the “number one enemy of productivity.” As improvements, Kaplan proposed Activity-Based Costing; Goldratt, throughput accounting.

In the 1990s, as Lean was gaining traction in the US, authors like Brian Maskell, Orrie Fiume and Jean Cunningham proposed various ideas under the label of “Lean Accounting.” I see Maskell as focusing on accounting for “Value Streams,” while Fiume and Cunningham emphasize reports that managers can understand without special training and that do not mislead them.

0.3. Management Accounting in TPS

Remarkably, none of this came out of Japan. The message I get from the TPS literature is that we should not pay too much attention to management accounting. Taiichi Ohno, for examples says “Costs do not exist to be calculated. Costs exist to be reduced.” Literally, it makes no sense. Costs are numbers, intended to quantify the consumption of resources for the purpose of supporting decisions. What Ohno is really saying is that we should not spend too much effort at getting the numbers “right,” but should instead focus on bringing them down.

In “Toyota Management System,” Yasuhiro Monden does not discuss management accounting as a function. He does have a chapter on Target Costing and Kaizen Costing, which, consistent with Ohno’s recommendation, covers how costs are used in managing the business but not how they are calculated. There is nothing, for example, on overhead allocation or on the handling of depreciation, that are at the core of management accounting in the US.

A 1988 article by Toshiro Hiromoto in the Harvard Business Review about Japanese Management Accounting extols it as “another hidden edge.” In particular, he thought it was a great idea for Hitachi to use management accounting as a tool to drive the implementation of its strategy. The example he quotes has the system apply overhead surcharges to products that use nonstandard components, as a means of discouraging this practice.

But it seems more like a misunderstanding of the whole point of management accounting, which is  provide objective information for management to set strategy. It is not intended to be an implementation tool. The article begs the question of what information Hitachi’s managers had used to determine that eliminating non-standard components was a strategy they should pursue.

This being said, looking at the record of the past decades, lack of sophistication in management accounting has not prevented Toyota from being successful. And where Japanese industry is struggling today, as in electronics, nobody is blaming management accounting.

The bottom line on Lean Accounting is that it is an approach developed in the US and not based on Toyota’s practices. It doesn’t mean there is anything wrong with it, only that it must be judged on its own merits and not presented as part of TPS.

0.4. The Economics of Lean

This is a broad subject that deserves more than a few sentences. Most attempts at quantifying the financial benefits of a successful Lean implementation are mistakenly focused on cost savings. For example, the point of quick setups is not to save costs but to enhance flexibility, which makes the company more attractive to customers.

You do many things to increase sales. You advertise, you sign up more distributors or dealers, you adjust prices, and you improve your manufacturing performance. Some of these actions may be effective and others not, and it is not easy to know which ones. Even when you know an action helps, you often cannot separate its impact from others’.

For example, if you advertise that you can deliver any custom configuration of your product in 5 days anywhere in the US, and sales go up by 15%, you know the market responded. But you still don’t know how much of the response was specifically due to the content of your message as opposed to the artistry of its presentation, or even just the amount you spent on advertising.

Focusing instead on cost savings may be easier, especially for accountants, but it is largely irrelevant.

Note: The featured image from AccountancyAge, 11/2/2011.

Perspectives on Standard Work

In the TPS Principles and Practice group on LinkedIn, I started a discussion by asking “What do we mean by ‘Standard Work’?” At that point, I saw it as whatever you do to ensure that same work is done the same way every time, regardless of time of day, production line, or factory; 96 comments later, my perspective has changed somewhat.

Ensuring consistency is, of course, essential but the tool to do it is Job Instruction (JI) rather than Standard Work. An operator’s complete job often involves multiple tasks, each of which has its own instructions. Standard Work builds on these instructions by specifying how these tasks are sequenced and combined to make effective use of people and equipment.

The main contributors were Len CanootSid JoynsonPeter WintonCasey NgDavid Hayden, MBAAchyut VaidyaEmmanuel JALLASEdward M. WhartonStephen DuquetteErik HagerJoachim KnufPaul Perry,  Cid LiSalvador D. Sanchez, Richard KunstAnders PenkerAndrew Williamson, and Steve Milner. The discussion also cited publications by Mike Rother, Art Smalley, John Shook, and Taiichi Ohno. If you want to see the whole discussion, please check it out on LinkedIn. This post is a synthesis , organized by topic and with illustrations added.

1. Why ask about Standard Work?

The House of Lean is a common metaphor. I use it sparingly, to make the point that the reason most Lean implementations in the US fail is that they are missing one of the two pillars:

House of Lean in Working with Machines

For this purpose, I don’t need to break down the details of what is in the Foundation or what the Goals are. Others provide many more details about the House of Lean, using it as a map of the whole system, with a prominent place given to something called “Standard Work” or “Standardized Work”:

As we can see, there is with variation in Standard Work  is supposed to mean. The Toyota description of Standard Work, for example, includes no reference to 5S or Visual Management, and explicitly excludes Job Instruction. The house on the right is from the Lean Enterprise Institute’s Lean Lexicon, and lists “Standard Work” and “Separating human work and machine work” as distinct entries but it is exactly what you accomplish with work combination charts, that are part of what Toyota calls Standard Work.

When you look it up on the Toyota Georgetown website or the LEI’s Lean Lexicon, it is about setting, for each operation, a takt time, work sequence, and required WIP, as expressed through a process capacity sheet, a work combination chart, and a work chart that is a layout diagram showing flows of parts and movements of operators between stations. (Click to see in full size.)

This is much more specific than what is meant by Standard Work is most discussions I have seen. I use capacity sheets, work combination charts and work station layout charts wherever you have to choreograph people working with machines, but I would not recommend them, for example, in manual assembly.

I have posted before about the work combination chart, as a powerful design tool for operator jobs, that also serves to communicate the sequence of tasks to operators, particularly new ones who are rotated into these positions. I see them as excellent tools, but it would not occur to me to label them “Standard Work,” because I don’t see any connection with the usual meaning of “standard.” I understand that “Standard Work” is an accurate translation of 標準作業 (Hyojun Sagyo), but I still don’t see a connection.

The following video clip, posted by JMA in Japan in 2009, shows what can be accomplished with work combination charts:


2. Standard Work and Process Stability

A process is stable if it can produce consistent output at a consistent pace. If it’s not stable, the first order of business is to stabilize it, but I don’t see standard work as the way to do it. You need to re-engineer the process to the point that its capability is no longer an issue and it is repeatable. Documentation and work instructions are an outcome of this effort, as needed to reduce the improved process to daily practice, but it is not the effort itself.

And the resulting documentation is not Standard Work. Standard Work, in the Toyota lexicon, is about takt time, work sequence, and required WIP, it doesn’t include process capability or even work instructions at the individual station level. It is only about the way you combine them in a line or a cell.

3. Is Standard Work the Best Known Way?

Most the Lean literature depicts Standard Work as an improvement backstop, a formalization of the improved process for the purpose of preventing backsliding. The following video is a quaint example of a PowerPoint animation used by a consultant to make that point. Note the yellow block preventing the wheel rolling back down:

Standard Work as “the best known way of doing the task” is the improvement backstop view, which I held without questioning it until I saw two articles disagreeing with it, and with each other, by Art Smalley .and Mike Rother. Mike Rother sees standard work as a target to shoot for rather than a backstop. Following is his rolling-wheel diagram:

Mike Rother's standard as a target condition

Mike Rother’s standard as a target condition

Yet another version was included in John Hunter’s review of Gemba Walkabout, and it shows standard work used to block progress instead of helping.


These rolling-uphill diagrams remind me of the myth of Sisyphus, as described by Albert Camus. Sisyphus was a man condemned by the Gods to roll a boulder uphill everyday only to see it roll back down and start over, for eternity. See the following rendition by Marcell Jankovics:

Even if Sisyphus had had a backstop, it does not strike me as a particularly attractive metaphor for Kaizen.

Art Smalley sees Standard Work as a point of reference against which to measure future improvement. Taiichi Ohno does not say much about it in “Toyota Production System,”  but in Workplace Management,  he writes:

“There is something called ‘Standard Work,’ but standards should be changing constantly. lnstead, if you think of the standard as the best you can do, it’s all over. The standard is only a baseline for doing further kaizen. lt is kai-aku if things get worse than now, and it is kaizen if things get better than now. Standards are set arbitrarily by humans, so how can they not change?

When creating Standard Work, it will be difficult to establish a standard if you are trying to achieve “the best way.” This is a big mistake. Document exactly what you are doing now. lf you make it better than now, it is kaizen. lf not, and you establish the best possible way, the motivation for kaizen will be gone.

That is why one way of motivating people to do kaizen is to create a poor standard. But don’t make it too bad. Without some standard, you can’t say “We made it better” because there is nothing to compare it to, so you must create a standard for comparison. Take that standard, and if the work is not easy to perform, give many suggestions and do kaizen.”

4. John Shook on Standard Work

John Shook published three e-Letters on the subject of Standard Work in October 2009, called “Five missing pieces in your standardized work.”

In Part 1, he describes the goal of having the same work done the same way everywhere as distinct from Standard Work. He calls it “commonization” as a translation of 横伝(Yokoten). Literally, Yokoten means “lateral transfer,” but it is meant about know-how, not people. You invented a better way to do a job, and you propagate it to everybody else who does the same job.

When he discusses the distinction between Standard Work and Work Standards, Shook includes under Work Standards not just the time a task is supposed to take but all its technical parameters, such as critical dimensions, tolerances, etc.

He describes Kaizen and Standard Work as two sides of the same coin. You can’t have Kaizen unless you have Standard Work as the basis for improvement, and a Kaizen project is not finished until its outcome is incorporated in Standard Work. But Standard Work as he describes it —  with work combination charts — is used almost nowhere in American plants that claim to practice Kaizen. This means that some of the following must be true:

  1. The definition of Standard Work is too narrow. The need to specify takt times, work sequence and standard WIP is general, but different tools can be used to do it in different types of plants. A work combination chart, for example, is of limited value in a manual assembly process.
  2. Most plants that claim to practice Kaizen really don’t. In Japan, Kaizen designates small improvements to work methods, conceived and executed by the people who do the work, and US-style “Kaizen Events” are not Kaizen at all. A plant may run 50 Kaizen events per year and still not practice Kaizen. The means of implementing Kaizen include suggestion systems, that exist in many plants with varying success, and small-group, circle activities, that, in the US, are only found in Japanese transplants. As “Quality circles,” in the US, they were a fad in the 1980s; as Jon Miller pointed out in Quality Digest in 2011, circles are still going strong in Japan and in the rest of Asia.
  3. Some Kaizen activity is possible without Standard Work. What you really cannot do without is some metrics of before-and-after performance for the area that is improved, and these may be measured without Standard Work being in place.

The bulk of Part 2 is an example from Shook’s own experience on the Toyota assembly line in Takaoka in 1984. In Part 3, he describes Standard Work through the Purpose, Process and People framework, which he calls 3P. I had heard the “3P” acronym used before, by Shingijutsu people as the “Production Preparation Process,” which is something completely different.

5. Standard Work versus Work Standards

John Shook gives the following as examples of Work Standards:

  • Assembly – apply xx pounds of torque
  • Processing – heat treat at xxx degrees for x hours
  • Healthcare – provide xx medication at xx dose
  • Coffee – xx seconds for an espresso shot
  • Journalism – a weekly column of xxx words
Frederick Taylor quote

Frederick Taylor quote

Last month, the Institute of Industrial Engineers (IIE) had a conference in Chicago on “Managing Work Standards.” It was exclusively about how long it takes to do work, not about what the work is. It is a sensitive topic because it is associated in the minds of production operators with Taylor’s “scientific management” and his determination to prevent operators from colluding to curtail output, which he called “soldiering.” For all his great contributions, respect for humanity was not Taylor’s strong suit. He probably would have said that this man should have borrowed money from his parents to start a business…

What we are doing when analyzing video recordings of operations is more in line with what Frank and Lillian Gilbreth did: observing processes in order to improve them. The difference in thinking is obvious from just viewing the films the Gilbreths made about bricklaying operations.

The Gilbreths were working to make the bricklayers’ job easier, not to make them exert more effort, but Taylor’s name is better known, and his legacy is a challenge to live down.

I think we need to improve the terminology. Having two different concepts called “Standard Work” and “Work Standards” is confusing, especially when Toyota uses “Work Standards” to mean something other than the IIE. Incidentally, it is confusing in Japanese too.

How about using “Work Instructions” for what Shook calls “Work Standards”?

6. What is the Scope of Standard Work?

What is the scope of Standard Work? I have seen described, I don’t remember where, as the process as seen through the eyes of a first-line manager — also currently known as production supervisor, group leader, or area coordinator, and formerly as foreman. This is a member of management, with direct responsibility for quality, cost, and delivery by a few teams of operators.

This person sees the work as a sequence of tasks to which operators are assigned and among which they rotate as needed. The technical and human unit processes at each station are the foundation on top of which the supervisor works. This would be why Standard Work is focused on takt time, work sequence and work combinations, as opposed to tolerances and job instruction (JI).

7. Standard Work, Yokoten, and Revision Management

Also, Standard Work comes in the form of documents that are seen on the shop floor and that people are expected to follow. This makes them official, with revision numbers and approval stamps by stakeholders. Revision management on Standard Work is a whole other topic that I have not seen discussed anywhere.

8. Is Standard Work a Proper Focus for a Project?

Standard Work is a 2nd tier tool, like Visual Management, meaning that it is part of every project but never the focus of a project in its own right. In a brownfield situation, making “Standard Work” a project would lead you to attempt the precise documentation of work methods that need to be changed anyway, which would not be terribly useful and could bog you down for so long that you never get to do anything else.

On the other hand, if you identify specific dysfunctions in a process and organize a project to fix them, then you want the new and better way to be documented in such a way that it can be propagated across shifts and to other shops that do the same work.

9. Is Every Problem a Deviation from a Standard?

Peter Winton feels strongly that it is. And this is about standards in general, not just Standard Work. If every problem is a deviation from standards, however, we have an easy way of solving all our problems: let us just scrap the standards… But it would not solve all our problems, would it?

It would solve some problems, because there are futile standards. As David Meier pointed out, when you set a standard, you create an opportunity for deviation, and the need to respond to these deviations. So don’t standardize what you don’t need to.

The absence of a standard can be a problem. I remember a 2-in binder of specs on how to inspect an aerospace part that did not actually contain objective criteria for rejecting a part.

More generally, Standard Work, Job Instructions, Acceptance Specs, etc. are documents that are necessary to ensure a consistent output but not sufficient to guarantee that products will work for customers.

Products that are perfect on our terms may still displease customers, because they are using them in ways we didn’t anticipate. That is a problem, but it is neither the lack of a standard nor a deviation from any standard.

From what you write, I assume that you consider a standard to be an explicit statement of what should be, whether it is expressed as “this bolt should be tightened to x foot-pounds of torque,” or “this bolt should be tightened until the nutrunner’s light goes green.”

There are problems that cannot be expressed as a deviation from standard. As we all know, the proof of a cake is in the eating, which means that it cannot be tested before leaving the pastry shop. You serve this untested cake to your guests and it’s awful. The taste of the cake, in Juran’s terms, is a true characteristic. It is really what you are after but, more often than not, it is something you don’t know how to measure, and you can’t set a standard for.

You can measure some substitute characteristics of the cake, like its diameter, sugar content, or fat content. For these substitute characteristics, you can have specs to deviate from, and, if a cake is out of spec, you know it’s bad. It is, however, possible for a cake to meet all the specs you have defined and still taste awful. Whatever standards you define work as a one-way filter. What they allow you to reject is defective, but you don’t know that what they let through is not.

Philip Crosby

Philip Crosby

Joseph M. Juran

Joseph M. Juran

This was the old debate between Philip Crosby, for whom quality was “compliance to requirements,” and Juran, for whom it was “the agreement of reality with expectancy.” These are different philosophies, leading to different practices. For engineering students, for example, the Crosby approach would equate scoring As on exams with being a good engineer; in the Juran approach, there is more to it.

10. Is it “Standard Work” or “Standardized Work”?

It is “Standardized Work” that is the questionable translation. The Japanese term is 標準作業 (Hyojun Sagyo). 標準 (Hyojun) means Standard and 作業 (Sagyo) means work. Google translates 標準作業 to Standard Work and Standard Work to 標準作業. If you translate “Standardized Work” into Japanese, you get 標準化された作業 (Hyojunka sareta sagyo) and more syllables on both sides. I prefer the shorter version.

The same concept is called “Standard Work” by Ohno, “Standard Operations” in the JMA’s ‘Kanban, Just-in-Time at Toyota” and in Monden’s “Toyota Production System,” and “Standardized Work” in the LEI Lean Lexicon and on the Toyota Georgetown website.

If the terms were intended to designate different things, they should be more distinctive. I actually don’t think either one makes much sense because they are too generic and not descriptive. If you hear “page scanner” for the first time, you guess accurately what it does; for “Standard Work” or “Standardized Work,” good luck! Everybody thinks they know what it means, but all interpret it differently, which does not help communication.

11. Standard Work for Leaders and Managers

David Hayden brought up the subject of Standard Work for leaders, managers, and engineers. I see all jobs as routinely involving a mix of the following:

  • Repetitive tasks
  • Planned responses to events
  • Decision making in the face of unplanned events.

For production operators, it is mostly repetitive taks; for CEOs, mostly decision making. Standard Work, if defined as the combination of takt time, work sequence and standard inventory, is only applicable to production operators. In a broader sense, it can be applied to all repetitive activities.

A team leader in a cell, for example, does production work for about 50% of the takt time, and, in addition, is responsible for

  1. Maintaining the pace.
  2. Relieving other team members as needed.
  3. Supplying materials and tools to other team members.
  4. Keeping records.
  5. Coordinating changeovers.
  6. Coordinating 5S at the end of the shift.

Planned responses are not Standard Work in the strict sense. As far as I know, within Toyota in the US, they are organized under “Change Point Management” (CPM). In Japanese, as Casey Ng pointed out, it is called 変化点管理 (Henkatenkanri). About the scope of CPM, he wrote:

“For change point management such as a change in takt time , the introduction of new members to a  line, preparation to shut down, startup after week-end, resuming production after a power failure, introducing a new product, changing of new version of parts etc. There are all sort of standards which may generally call Standard Work.”

12. Standard Work and Project Management

Len Canoot asked whether the elements of Standard Work were translatable to project management.

The time it takes to do the work is the process time or the cycle time, not the takt time. In a line that works at a steady pace, the takt time is the interval between consecutive unit completions in order to meet the schedule within the work time available.

Does it translate to projects? It depends what kind of projects. If all your projects are “Kaizen events,” each one takes 11 weeks: 6 weeks of preparation, 1 week of focused activity, and 5 weeks of follow-up. It is a standard process, and you can to run them at fixed intervals in different areas of your plant. Most projects, however, are not reducible to this kind of cookie-cutter approach.

In a more general setting, there are tools you can use to manage a flow of projects, like capping the number in progress, so that participants’ attention is not spread too thin. At the very least, all your projects are either waiting to start, in progress, or finished. Often, however, they all go through a more detailed, common sequence of phases through which you can track them, even though the work required for a project through a given phase may vary.

Does that add up to standard work for projects?

13. Standard Work versus Standards in general

About the usual meaning of standard, this is what Wikipedia says about technical standards:

“A technical standard is an established norm or requirement in regard to technical systems. It is usually a formal document that establishes uniform engineering or technical criteria, methods, processes and practices.”

That covers the metric system and the internet protocols. Wikipedia also says the following about Standard Operating Procedures (SOP):

“In clinical research, the International Conference on Harmonisation (ICH) defines SOPs as ‘detailed, written instructions to achieve uniformity of the performance of a specific function’.”

When I see “detailed, written instructions,” it makes me think of the victorian-novel sized instruction binders that sit on shelves in many plants, unread, dusty, and full of obsolete information. Standard is also used in many other ways.

14. Standard Work and Changes in Takt Time

Anders Penker brought up the issue of the effect of changes in pace on Standard Work. One key reason you set up U-shaped cells, with the operator work area inside, is staffing flexibility. If it takes, say, 5 operators to operate at capacity, you can operate with 3, 2 or 1 operator at reduced rates, as seen below:

Takt time change in cells

Of course, your work chart and work combination chart for 6 operators are not applicable when you only have 4 or 3. But operating at a reduce pace with fewer people is something you can anticipate and plan for. You should have Standard Work ready for these circumstances, and post it as needed.

And there are circumstances where these charts are not applicable, for example when you apply the bucket-brigade method to make custom-configured products. But that is a different topic.

Silos, “Value Streams,” and Lean Implementation

The on-line chatter about Lean is all about how you need to break down functional departments — or silos — and organize the company around “Value Streams” that encompass all the resources needed to fulfill orders for a product or product family, and are close cousins of BPR’s business processes and Wickham Skinner’s focused factories. When discussing implementation, however, the same value stream boosters/silo busters recommend that you start by setting up a “Kaizen Promotion Office” or “Lean Department.” This reminds me of the 1980s BBC series Yes Minister, in which an effort to streamline government starts with the creation of a new “Ministry of Administrative Affairs” and the hiring of 25,000 more civil servants to do the streamlining.

While it is ironic to create a new functional department while talking value streams, it reflects a reality: the notion of organizing everything by value stream is simplistic. As discussed in my comments on Deming’s exhortation to break down barriers between departments, there are many activities in a manufacturing organization that we cannot or should not distribute among value streams, including the following:

  1. Processes like heat treatment, painting or plating that we have to operate as common services performed on monuments for multiple value streams because we technically do not know how to execute them on smaller machines that can be dedicated by production lines.
  2. Support services like maintenance that require a minimum number of members of members for at least one to be available when called. If you have 20 technicians in a central maintenance department that are busy 80% of the time, then at least one will be available if a machine breaks 1-.8^{20}=99\% of the time. If you split this department into 4 groups of 5 technicians each assigned to a value stream, then, if a machine breaks down within any value stream technician availability will be reduced to an unacceptably low $latex 1-.8^{5}=67\%$ of the time.
  3. Support services that deal with external entities on behalf of the whole company or plant, like Quality or Safety for certification, or Shipping and Receiving with truckers.
  4. Support services whose job it is to maintain a common environment for operations, such as technical data management or IT.

As for the Kaizen Promotion Office or Lean Department, mission creep all too often takes it from a feasible facilitation and communication role to a direct implementation role, which is hopeless because:

  1. The operations groups have no ownership of the changes made by the Lean Department, do not understand them, and frequently reverse them as soon as they have a chance.
  2. The Lean Department cannot be large enough to have the capacity to do everything that is needed.

For the changes to happen and to stick, there is no alternative to leadership from within the organizations responsible for the target operations and participation by individuals who are directly affected.

Metrics in Lean – Chart junk in performance boards and presentations

Manufacturing professionals who read Edward Tufte‘s books on visualization may be stunned to discover that their 3D pie charts, stacked bar charts, and green safety crosses are chart junk. These charts are common, both on shop floor performance boards and in management presentations. But they are information-poor, and their decorative elements distract, confuse, and occasionally mislead.  The purpose of plotting is not to dazzle, but to discover patterns, understand the underlying phenomena, and communicate with people whose livelihood is affected by these patterns.

For details, see:

1. Pabulum pies

Following is an example of what Microsoft Support considers to be a pie chart with a “spiffy, professional look,” and a few comments about it.
This chart not only uses buckets of ink to display five data points, it also violates other good design principles:
  1. The legends are remote from their objects, forcing the reader to look in two places to understand each item.
  2. The chart begs a question that it could answer but doesn’t: percentages of what total amount?
  3. The shadow and the light reflection convey no information.
  4. The 3D effect makes the 21% slice appear larger than it actually is.
If we put the legends on their objects, add the total amount as a subtitle, get rid of the vinyl-cushion look, and take a top view so that the wedge sizes are not distorted, we get the following:
Microsoft pie chart example - improved

One feature that now stands out is that the wedges are ordered by decreasing size, except for the last two: “Desserts” is larger than “Beverages” but comes after. It is the example Microsoft uses for training and they don’t explain the sequence. While this chart is an improvement over the previous version, but do we actually need it? A picture may be worth a 1,000 words but, if what you have to say fits in 10 words, a picture may be overkill. Compare the pie chart with the following table of the data in it:

Microsoft pie chart example - data table

If you compare a pie chart with a sorted table of the data in it, you see that the chart uses orders of magnitude more ink in multiple colors than the table, without telling you anything that isn’t already obvious from the table. In Edward Tufte’s book, it makes it the kind of chart junk that you should banish from your materials.

3D pie charts are actually worse, because they are not only useless but misleading in that the perspective distorts the apparent size of the wedges. This is an illustration of another of Tufte’s principles, that a graphic display should not have more dimensions than the data. In 3D pie charts, the height of the pie is meaningless. The only circumstance in which showing pie thicknesses would be useful if when showing several pie charts side by side, and both pie diameter and thickness would represent parameters of each pie.

2. Stacks astray

Vertical bar charts, also known as column charts, usually serve to show the evolution of a quantity for which data is available for a sequence of periods. It is used when there are few periods and interpolation between periods does not make sense. If you plot daily sales over a year in a vertical bar chart, the columns will be so densely packed that you will only see the line formed by their tops anyway, and you might as well just use a line plot.
Bar charts - Daily sales as bar versus line plot
On the other hand, if you are plotting monthly sales, you use vertical bars, because interpolation between two points does not give you meaningful intermediate numbers. If you plotted temperature readings taken at fixed intervals, for example on a hospital patient, you would interpolate to estimate the temperature between readings, and therefore you use a line plot rather than vertical bars.
Bar chart of monthly sales

Bar chart of monthly sales

There is nothing objectionable to the ordinary, vertical bar chart. It is simple to produce and easy to read, and provides information that is not obvious in a table of numbers. Stacked bar charts, however, are another matter. They attempt to show the evolution over multiple periods of both an aggregate quantity and its breakdown in multiple categories, and do a poor job of it, especially when a spurious 3rd dimension is added for decoration.
We can retun to the Microsoft support web site for a tutorial on stacked bar charts. The building-block look of the Microsoft example may be appropriate for an audience of preschoolers.

Another oddity of the Microsoft example is that it does not follow the convention by which vertical bars are used when the categories on the x-axis represent consecutive time buckets. When they are not ordered categories, you usually prefer horizontal bars, not only because we are used to the x-axis representing time in performance charts but also because this chart has horizontal category labels, readable without tilting your head.

Based on the preceding section, the first question we might ask is whether we need graphics at all, and the answer is yes. When we take a look at the data in table format, even though there are only 16 points, no pattern is immediately apparent:

Stacked bar Microsoft example source table

If we forget the spurious 3rd dimension and the building-block look, and toggle the axes so that the x-axis represents time, we get the following:

Stacked bar Microsoft example 3D removed

From this chart, it is obvious that total sales collapsed from the 2nd to the 4th quarter; it was not obvious from either the table of number or the 3D chart Microsoft presented as a model. Even in this form, however, the stacked bar chart is ineffective at answering the immediate follow-up question of whether the decline in sales was more pronounced in some regions than others. For example, if we try to isolate the Northeast on this chart, we find bars floating at different heights. On the other hand, the answers become visible if you de-stack the bars, as in the following:

Stacked bar Microsoft example unstacked

For example, you can see that sales actually grew from the 1st to the 2nd quarter in both Eastern regions, while declining throughout the year everywhere else. Yes, it takes up more space than the stacked bar, but is that really a concern when, even in a simple example like this one, it lets you see better?

3. Safety crosses to bear

The safety performance of a plant matters, and not only to the potential victims of accidents. Injuries must be rare events, occur with ever decreasing frequency, and have both immediate countermeasures and permanent solutions to prevent recurrence. In light of this, what kind of graphic summaries would be appropriate to represent the safety performance of a whole plant, or of a shop within it?

A common metric is the number of consecutive days without a lost-time accident. It is a relevant measure, but imperfect in that its use has been known to lead thoughtless managers to blame victims for hurting their department’s performance and pressure them not to report injuries. It is also necessary to show detailed information about each accident, and to categorize injuries, for example, in terms of whether they affected hands, shoulders, feet, etc. and where exactly they occurred.

In light of these considerations, what is industry using? In the US, the National Safety Council uses a green cross in its logo and awards a Green Cross for Safety Medal to one company each year. That makes the green cross a symbol of safety, and it has motivated some to make it the basis of a safety performance tracking chart. You start with a cross shape subdivided into rectangles numbered 1 through 31 and, on each day of the month, you place a magnet on the corresponding spot, which is green if nothing happened, yellow if a minor incident happened, and red for a lost time accident.

It is difficult to think of the use of a chart in the form of a symbol of safety as anything but a gimmick. The green cross has no connection with any requirements we can think of for a chart of safety performance. It does not make it particularly easy to count consecutive days without incident, nor does it bear any information about the nature or location of the accidents. The green cross shape evokes safety but has nothing to do with the key questions about employee safety.

To visualize a sequence of rare events, a technique that comes to mind is the timeline. You arrange events around a centerline that shows the passage of time, as in the following example that summarizes 10 years of the history of the iPhone.

Timeline 10 years of iPhone history

Safety-related events in a factory do not need a 10-year timeline, but possibly a six-month timeline along the following model:

Timeline of sports events january-june 2010

Note that the time between events is immediately visible, and that each event has some explanation and photographic documentation. For location information, you can pin injury locations on an outline of a human body and a map of the shop floor. While these may not be pleasant charts to consider, they are a means of starting a conversation in the team on what the safety issues are and the means of preventing injuries.

There are, of course, other safety issues, like repetitive stress, that are not associated with discrete events and do not appear on these charts, but they do not appear on the green crosses either.

4. Recommendation on performance board design

Performance boards come in all sorts of shapes, as in the following examples:

As a performance board for a shop floor team, I recommend the following template:

Performance board template

Performance board template

This template has one column per dimension of performance and one row for each type of information, as follows:

  1. The top row is for the latest news, what happened this shift or last.
  2. The second row is for historical trends in aggregate performance.
  3. The third row is for a breakdown of the aggregate into its constituent parts, such as the most common injuries, defect categories, most frequently made products, or the employee skills matrix.
  4. The bottom is about actions or projects in progress in each of these areas.

Deming’s Point 10 of 14 – Eliminate slogans and exhortations

Deming’s full statement is as follows:

Eliminate slogans, exhortations, and targets for the work force asking for zero defects and new levels of productivity. Such exhortations only create adversarial relationships, as the bulk of the causes of low quality and low productivity belong to the system and thus lie beyond the power of the work force.

Ben Hamper’s Rivethead

This point reminds me of Howie Makem, the quality cat lampooned by Ben Hamper in Rivethead in 1986, about the same time Deming’s Out of the Crisis was published. At the time, Ben Hamper was a riveter at GM’s Truck plant in Flint, MI, who could describe his shop floor experience with the wit of a Tom Wolfe. Rivethead was originally a column in Michael Moore’s Flint Voice, later edited into a book.

According to Hamper, the management of the plant had decided that what it needed to improve quality was a mascot for workers to rally around, and organized a naming contest, of which “Howie Makem” was the winning entry. The mascot then materialized as a man in a cat suit with a large Q embroidered on a red cape walking the floor and exhorting operators to improve quality amid jeers, catcalls and the occasional bolt throw. Howie Makem is one of the few artifacts of which no picture can be found on Google, which is why I had to draw it from Hamper’s description.

Spending time and money on slogans, mascots, banners and monogrammed shirts or mugs is predicated on the assumptions (1) that quality and productivity problems are primarily due to lack of motivation in shop floor operators and (2) that it can be changed by the same kind of marketing campaign that works for selling detergents. Deming’s and Hamper’s point is that it is counterproductive and that these assumptions are false.

The key points that I see about appropriate public relations and communications around Lean are as follows:

0.1. Do it first, play it back later

Improvement does need marketing and promotion inside the company, to customers, and to suppliers, but not at the start of the effort, and not in this form.

The beginning of an improvement program like Lean transformation is when it is most likely to fail. At that time, the organization, from management to line workers, has everything to learn about its technical and managerial content, as well as the art of implementing it. It is then that they will make the most mistakes and therefore least need publicity. The first pilot projects only need to be known and understood by those who are directly involved, and should not be announced upfront with a marching band at an all-hands meeting. You are much better off trumpeting results once the projects are successes that can inspire others. And even then, it is not done with slogans but by testimonials of participants, demonstrating the improvements directly on the floor or in video recordings.

With outsiders as well, you do it first and play it back later. You don’t announce what you are going to do, but, once it is done, you make it a field trip destination for local schoolchildren as well as other industrial tourists.

0.2. Car companies and public relations on manufacturing

Toyota plants have visitor centers with posters on the products and cartoons explaining the production system to children and have a whole staff of professional tour guides taking groups on a set path through the plant, wearing headsets to hear the explanations. These tours are part of public relations and not given by retirees, as is the case at many other companies.


Porsche Leipzig

Porsche in Leipzig charges customers €1,000 extra to spend a day at the plant to pick up their Panameras or Cayennes, during which they get a tour of the shop floor featuring their version of Lean, a lunch at top of the visitor center, and an hour with a driving pro on the test track to learn how best to drive their new car in various conditions.A striking feature of this plant site, is that it is dominated by the round, inverted diamond shape of the visitor center, on the top left of the photogaph, between the test track on the left and the production shops on the right.


VW transparent assembly plant in Dresden

This is part of a new marketing trend in Germany, where, rather than hide plants away, you locate the cleanest, most automated and most spectacular processes where your customers, or even the public at large, can see them. In this spirit, Volkswagen has located a plant downtown Dresden, with glass walls for passersby to see the final assembly of cars.

Motorcycle homecoming at Honda in Marysville in 2005

Motorcycle homecoming, Honda Marysville, 2005

Honda pioneered a different form of promotion of its manufacturing system to end users with its homecomings at the Honda motorcycle plant in Marysville, OH, where, once a year, they hosted bikers who see the production lines and meet the operators who built their bikes. The same approach was later emulated by the now defunct Saturn division of GM.

Companies in other industries rarely go this far, particularly when their products do not excite the public’s imagination. Bart Simpson’s class goes on a field trip to a box factory, which does not generate much enthusiasm.

0.3. Promotion of Lean efforts by component suppliers

If you make components to sell to OEMs rather than to consumers, the promotion of your Lean programs takes a different form, with customers sending teams of auditors to assess whether you are “Lean enough” to do business with, and they may send you supplier support engineers to help you implement Lean to their satisfaction. This means that you must present your plant in a way that allows the auditors to check all the marks needed to give you the right score, even if it means setting up a Potemkin village with tools that you don’t think are essential to your business.

Working with your customers’ supplier support organization — or supporting your own suppliers — is a different process, requiring a deeper level of involvement, and it is not a matter of public relations for either side, and should not be treated as one. The customer provides free consulting to help the supplier increase productivity and improve quality. In exchange, the supplier reduces prices by a fixed ratio every year, calculated so that the improvements are to economic benefit of both sides. The customer pays less, while the supplier makes more profits. It is a win-win, but not an easy system to set up and operate. It involves top management, engineering on both sides, purchasing on the customer side, and customer service on the supplier side, and it is not run by Public Relations.

Deming’s Point 9 of 14 – Break down barriers between departments

(Featured image from the  Bureaucracy game, by Douglas Adams)

Deming’s complete statement of Point 9 is as follows:

“Break down barriers between departments. People in research, design, sales, and production must work as a team, to foresee problems in production and in use that may be encountered with the product or service.”

Within a large organization, it is common for departments to work at cross purposes. Each department is a functional silo, working towards goals that may be inconsistent with the interests of the whole. Deming gives many examples of disasters that occur as a consequence, and exhorts his readers to break down the barriers to keep them from happening. As with his other points, he makes no recommendation on how to accomplish this.

Let us examine several approaches that have been tried, and the issues that organizations encountered when they did:

0.1. Eliminating silos in the organization

This is not a problem for small companies. As long as the entire management team fits within a small conference room, there are few opportunities to erect barriers. In a large company where it is a problem, the most obvious solution is to organize by what is variously called business teams, business processes, value streams, or focused factories.

You dissolve the functional departments and organize multifunction teams that bring all the required talent to bear on the core activities. In a manufacturing company, for example, all the resources needed to make a family of products from start to finish — including engineers, maintenance and quality technicians, schedulers, etc. — report to one “value stream manager,” and there cannot be barriers between silos because there are no silos.

It’s like the Mission Impossible TV series, with the disguise specialist and the explosives expert working together towards a common goal, as opposed to being in separate facilities and exchanging service requests in triplicate. This is a popular picture in the US and the approach is often used in a variety of contexts, such as emergency response, as in Apollo 13, or product development, for Data General’s MV-8000 computer in 1980 in Tracy Kidder’s The Soul of a New Machine, or the 1996 Taurus at Ford in Mary Walton’s Car.

The movie Apollo 13 shows a seemingly too-good-to-be-true team that is thrown together to find a way to fit the square connector of the command module air scrubber to the round hole used on the lunar module, using nothing but the odds and ends available to the astronauts on the crippled spacecraft. But the story is true, and we have a picture of the actual device the astronauts built.

This was the philosophy of Business Process Reengineering (BPR). Each business was to be broken down into processes turning some input into an externally visible output. Manufacturing, in BPR, did not qualify as a process. Instead, it was subsumed into the order-fulfillment process.

0.2. Making functional departments work

But it is not a panacea. The development of the 1996 Taurus took 30 months, and it was a major improvement over previous products at Ford, but still not down to the 24 months used at Toyota for the Rav4, and Toyota uses a traditional structure with functional departments communicating through memos.

In addition, according to Mary Walton, Ford’s integrated, collocated team made design decisions that made manufacturing more difficult. She explains in particular that the sculptured shape of the side panels made them more difficult to stamp, and this happened even though manufacturing was represented in the team. As a work of art, the 1996 Taurus was stunning. As a commercial product, however, it was lackluster, losing the previous versions’ bestseller status in the US market to the more “boring” Honda Accord and Toyota Camry in 1997.

The reality is that organization structure does not determine outcomes. The caliber of the individuals, their motivations for the roles they are playing, and their interaction protocols are at least as important. In their July, 1998 Harvard Business Review article , D.K.Sobek, J. Liker, and A.C. Ward listed the following practices as key to Toyota’s performance in product development:

  1. Written communication with single-sheet A3 reports in standard formats.
  2. Engineering supervision by practicing, hands-on engineers.
  3. A chief engineer (shusa, or 主査) for each project who is an experienced designer with a proven ability to integrate different technologies into a product. The shusa has a team of 5 to 15 members coordinating the work of hundreds who remain in functional departments.
  4. Engineers who develop their skills through on-the-job training, mentoring, and rotation within their functional department, with senior managers rotating between departments.
  5. High-level project plans with a small number of milestones, giving each department flexibility on detailed tasks.
  6. Checklists of design standards embodying the lessons learned in previous projects.

0.3. Obstacles to organization by process or value stream

The Toyota example is about product development. But what about other activities like operations? When you attempt to organize everything by business process, or by value stream, in most cases you encounter some functional departments that you technically cannot or should not break up.

Most machine shops have a central heat treatment facility. Induction hardening can, for some work, distribute heat treatment among different production lines and break down the “heat treat silo,” but a given shop may make products to which it is not applicable, its customers may not approve the process, or it may not have the skills or resources to implement it. Electroplating and painting commonly are similar challenges. As a result, the plant ends up with a few common services organized as functional departments along with lines that take a family of products through a sequence of operations.

Among support functions, the picture is also mixed. Production scheduling at the detailed level, for example, works better when the schedulers work directly for the manager of a production line than in a central department, because local scheduling is a simpler problem and the relevant specifics of machine behaviors are more accessible. On the other hand, breaking down a maintenance department and making the technicians report to production managers may not enhance their responsiveness when, for example, the group assigned to a line is short of the critical mass needed to have at least one technician standing by for the next emergency.

Other departments remain organized centrally because of the information they have access to, like Human Resources, Accounting, or Technical Data Management; others, because of external entities they deal with, like Shipping and Receiving.

0.4. Skills maintenance, continuing education and career planning

When breaking down a functional department and reassigning its members to teams organized around processes, we also need to consider how it affects the people to whom we do it. Professionals like medical doctors or lawyers work for clients who have little or no knowledge of their specialties, but it is then up to them to decide how much of their revenue to spend or maintaining their skills. They choose which magazines tp subscribe to and which conferences to attend, without asking anybody’s permission.

An engineer reporting to a production manager also works for one “client” who does not have the same expertise, but as an employee. If this engineer wants to attend a conference, the first step is to get approval for the time and money it will consume, from a manager with no knowledge of whether it is a good idea.

In the long term, what career does this engineer have to look forward to? The manager needs the engineer’s skills here and now but is ill equipped to provide guidance, compared to an engineering manager whose background and experience are in the same field.

For this reason, some companies have adopted matrix organizations, in which specialists report “solid-line” to a process owner who needs their skills in operations or on projects, and “dotted-line” to a functional manager for skills maintenance and career development. In a diagram, as follows, this structure looks simple and attractive:In reality, of course, it is a more complex form of organization than a simple hierarchy, and conducive to all sorts of tensions regarding authority and responsibility.

0.5. Project transitions

Project work — like product development, new product introduction, or new plant setup — differs from operations in that it ends when a goal is reached, which may be a working prototype, a target takt time in production for the new product, or for the new plant. At that point, the teams are disbanded and their members move on.

This is a particularly sensitive transition to manage when you collocate a multifunction project team in one big room, because its members bond both with the project and with each other, and receive the ending like a psychological blow on the scale of the loss of a family member. This is another reason why they need to retain a connection with their functional peers.

0.6. Conclusions

Breaking down barriers between departments for the greater good of the organization as a whole is a worthy goal, that high-level managers have been pursuing since, at least, the Roman empire. There is no simple recipe. The approaches followed by successful organizations have been subtle, nuanced, and fitted to their purposes.

Deming’s Point 4 of 14 – End the practice of awarding business on the basis of a price tag…

(Picture from TYWKIWDBI)

The complete wording of Deming’s Point 4 is as follows:

“End the practice of awarding business on the basis of a price tag. Instead, minimize total cost. Move toward a single supplier for any one item, on a long-term relationship of loyalty and trust.”

Today, you will encounter no one involved with supply chain management who would argue against the idea of basing purchasing decisions on the total cost of having the item on hand when needed for production and developing collaborative relationships with suppliers. The idea of single-sourcing every item, on the other hand, makes many managers nervous, but, without such a committed relationship, you cannot have information exchange at the depth required for collaboration to pay off.

As of 2012, however, very few companies have followed through on this recommendation. What we have seen instead in the past 20 years instead is “We’ll skip Lean and go straight to China,” based exclusively on temporarily cheap labor, without due consideration to local infrastructure, quality and productivity issues, and logistics. Companies that are “reshoring” after being burned at this now have an opportunity to implement this most specific and least controversial of Deming’s 14 points.

It breaks down into the following specific recommendations on what is now called supply chain management:

  1. End the practice of awarding business on the basis of a price tag.
  2. Minimize total cost.
  3. Move toward a single supplier for any one item.
  4. Develop long-term relationships of loyalty and trust with suppliers.

1. Stop awarding business on the basis of a price tag

In this area,  companies don’t behave like individuals. Whether you buy food, clothing, household appliances, or the services of a plumber, you don’t systematically choose the lowest price. Like the astronaut on the launch pad, you do not want every part in the rocket to have been made by the lowest bidder. Even if you are hunting for bargains, you also consider quality, delivery, and the availability of support. You willingly pay more for appliances that are reputed for working well, lasting long, operating quietly, match the design of your house, and have spare parts and service readily available.

In principle, a company’s purchasing agents should think the same way. When they don’t, it is because they are evaluated on the prices they are able to negotiate and because they are not familiar with the actual use of the materials or equipment they buy. If you hired a third party to do your shopping, with instructions to find the lowest prices, you are unlikely to be happy with the results or even to same money over time.

Because they don’t use what they buy, purchasers rely on specs to decide whether a supplier’s product meets the company’s needs. As Deming points out, however, conformance to specs is never synonymous with fitness for use, no matter how carefully the specs are written. Specs only work as a one-way filter; if a product is out of spec, you know you can’t use it, but, if it is within specs, it does not guarantee that you can. Juran distinguished between true and substitute characteristics. The true characteristics are what you are really after, like the taste of a cake. Unfortunately, you cannot verify it without eating the cake, so you use substitute characteristics that you can measure, like the cake’s diameter or the sugar content of the ingredients. If they are out of specs, you know there is something wrong with the cake, but they can all be within spec and the cake still taste awful.

Relying on specs in purchasing is therefore taking necessary conditions and treating them as sufficient. But how do you avoid doing this? Deming does not say. I recommend the following:

  1. Avoid perverse incentives. Use metrics for purchasing that do not overemphasize the price.
  2. Implement Lean supply chain management. It is a broader subject than just buying based on price, but it provides a context for a more balanced approach to evaluating suppliers.
  3. Rotate professionals in and out of Purchasing. This means treating purchasing as a skill employees should have rather than a career. If you have people in Purchasing who have previously worked in Production or Engineering, they will have a better understanding of the issues.
  4. Give end users a voice in Purchasing. Purchasing should not have the authority to switch suppliers without the approval of those who consume the materials or use the equipment.

2. Minimize total cost

For manufacturing, it means considering everything it takes to have good materials within arm’s reach of the production operator for as long as the line is running on this product, as opposed to the price on a purchase order. For equipment, it means looking at the total cost of ownership (TCO), also a term that was introduced after Out of the Crisis came out.

The only issue Deming raises is that of quality, but it is not the only one, particularly when you consider switching from a  supplier located 10 miles from your plant to one that is 6,000 miles and 10 time zones away in an unfamiliar country. You have to consider transportation, longer lead times, communications and travel.

Furthermore, discussing cost and quality in the same breath leads naturally to thinking about what the literature calls “cost of quality.” The literature on quality defines this cost as the sum of the direct costs of failure, appraisal and repair, and omits the impact of quality on sales, as being too “controversial” and difficult to measure. This “cost of quality,” however, is the tip of the iceberg; it grossly underestimates the business consequences of quality problems, as shown, for example by the Firestone tread separation issue in 2000 or Toyota sticky accelerator pedals in 2010. A car maker’s reputation for quality is its crown jewels, and the answer on how much effort it should put into nurturing is is whatever it takes.

While transportation costs are relatively easy to calculate, the cost of expanding lead times from days to weeks or months is, in some cases, much larger than the cost of having inventory in transit. For example, toys sold in the US during each Christmas season are made in China the previous summer, but you cannot tell bestsellers from duds until late in the fall, by which time there is nothing you can do to adjust the supply.

To follow Deming’s recommendation here, you consider not the unit price of the item but all the outflows of funds generated by the decision to buy it for a given supplier for as long as you intend to do it, knowing that this may vary from a few months for fashion-related items to several decades for airplane parts. The question is not the price of one unit but, for example, what it takes to make, say, 1,000 usable units available on your production line every day for the next four years. And you have to write at least a best-case, worst-case, and most likely scenarios of how it may unfold in terms of volumes, quality and delivery performance, and  technical support of the supplier. Each scenario results in cash flow schedules that can be compared.

Such an analysis cannot be done without making assumptions about product life, demand, and supplier capabilities. It is more complex than picking the lowest bidder, but the stakes are high.

3. Move toward a single supplier for any one item

What happens when your single supplier fails? It happened to Toyota with the Aishin Seiki fire of February, 1997. The plant was Toyota’s single source of proportioning valves for Toyota in Japan. Toyota’s factories shut down within four hours of the fire, the supplier network was mobilized, production was restarted within a week, and was back to full volume in 6 weeks.  In the Japanese press, the fire was initially viewed as a failure of Toyota’s system; by the time it ended, it was a vindication of it.

If you buy thousands of items, even with a single source for each, you will have hundreds of suppliers. If you have a policy of having at least two sources for each item, you will have even more suppliers and more complicated relationships to manage. Deming emphasizes the impact on quality, but it touches in fact every aspect of supplier relations. Juggling multiple suppliers for each item is playing the field; having a single source, a monogamous relationship.

If, for each item, you have a single source for whom you are a major customer, your plan for dealing with emergencies like the Aisin Seiki fire is to rely of the strength of your supplier network to come up with an appropriate response. The Wall Street Journal article about the Aisin Seiki fire in May, 1997 described the response of Toyota suppliers as the manufacturing equivalent of an  Amish barn raising.

Sudden surges in demand are not an issue in car manufacturing, but they are in other industries, like semiconductor production equipment. If you are a machine shop making components for this industry, you may see demand doubling overnight simply because  one semiconductor company placed a big order for machines in a new wafer fab. You know that sudden changes in the economy may cause this order to be cancelled, and you cannot count of other orders filling up you slack capacity once this order is filled. In this case, rather than investing in additional equipment that is unlikely to be permanently needed, suppliers have been known to make second-sourcing agreements with competitors to provide surge capacity. One consequence of such arrangements is that the parts arriving at the customer plant may come from different suppliers. From the customer’s perspective, however, it is still a single-sourcing arrangement,  because the primary supplier remains responsible for quality and delivery.

4. Develop long-term relationships with suppliers

A six-year contract representing 30% of your sales to be a customer’s sole supplier of a component sets the stage for a different working relationship than a one-year contract representing 10% of your sales, in which you are one of a stable of suppliers among which the customer splits the demand. Exclusive, long-term relationships are clearly a required foundation for the collaboration that the entire literature on supply chain management agrees should take place between suppliers and customers, but generally doesn’t.

“Arms around” is better for both sides than “arm’s length” and adversarial. So why is it so rare, and what can we do to make it more common? The abundant literature on supply chain management fails to see what I think is the elephant in the room: unlike a plant, a supply chain is ruled by the interaction of multiple, independent economic agents. This is discussed in Chapter 19 of Lean Logistics (pp. 341-352). The summary is as follows:

In the lean supply chain, the traditionally adversarial, arm’s length relationship between supplier and customer makes way for a collaborative approach, centered on long-term single-sourcing agreements, and extensive exchanges of business information and technical know-how. This approach increases the total payoff of the relationship, but transitioning to it is difficult because it requires behavior changes on both sides.
Sustaining it over time also requires management to consistently forgo the short-term windfalls that can be reaped through a unilateral return to the adversarial approach. That supplier and customer should collaborate to increase the total payoff does not prevent each one from negotiating aggressively with the other on sharing this payoff.

Once you acknowledge that a collaborative relationship takes a long time to build and are easily destroyed by either side, you can manage it accordingly and give it the attention it requires.

What to look for on a gemba walk

Gemba (現場) means actual place. As consultants, we ask clients to show us their gemba, and we exhort their managers to do it routinely. But we must have a clear idea of why we should go to the gemba and what to do once we are in it. In Manufacturing, the gemba is the production shop floor.

For a consultant, the point of walking through a factory shop floor is to learn about its current state, complete through direct observation what could not be known through previously received written or oral input, and to validate or refute this input. For a manager, making daily rounds through the shop floor is different, and involves two-way communication. The manager’s presence, body language, and attire are by themselves a message to the work force. Being watched by everybody, the manager can listen and ask questions, but must be cautious not to give instructions to operators over the heads of supervisors.

From documents received ahead of time or personal communication, the consultant might know that the plant is using dispatch lists from an ERP system to schedule production. On the shop floor, these dispatch lists and the way they are used would be visible. Manual annotations could reveal that the tasks are not done in the recommended sequence, and a supervisor could explain that this is due to setups or missing parts. In other words, you don’t go to the shop floor to find out what the intended scheduling system is, but to find out how work is actually sequenced and what relationship it has with the scheduling system output. A manager walking with the consultant would make a note of the situation, and follow up on it later with supervisors and Production Control.

Shop floor observations include the overall design of the plant for production and internal logistics, as well as details that reveal how it is operated. You can tell whether it is a job-shop, a flow line, or a collection of flow lines. You can tell whether the flow of materials is visible, what kind of equipment is used for materials handling, and how much of the floor is used for warehousing versus production. When you zoom in on individual stations, you can assess the level of automation and the attention that has been paid to the design of operator jobs. You can also check out the accuracy of the signage, the presence and use of andons, mistake-proofing devices, production monitors, and team performance  boards.

It is quite possible to walk through the aisles and not notice that the plant is anything but a tight ship. The key to actually seeing is to not just watch but instead act. This activity yields information both directly and indirectly. Several tools are available to help you see better, some of which require more than a quick visit. They include the following:

  1. Using a Seven Wastes Checklist. The list of seven wastes helps you identify occurrences of them, whether you keep in on a paper checklist or in your mind.
  2. Following the flow. Pretending you are a work piece and following the process backwards from the end to the beginning,  noting where and how many times it waits for transportation or processing, how operators perceive upstream and downstream colleagues, the tools,  fixtures and storage devices  used at each operation.
  3. Counting. You emulate Sesame Street’s “The Count”  and start counting people, machines, parts or fixtures. That’s how you may notice that 20% of the people are walking in the aisles rather than tending to their machines. You ask a few questions and find out that half of those 20% are going to or returning from the tool crib. You have not only discovered that the plant uses a wasteful method for distributing tools, but you also have a ballpark estimate of the productivity improvements at stake in setting up tool pickup and delivery milk runs. Thus the simple act of counting people has led you to discover a pattern of wasteful operation, which you will then recognize immediately elsewhere. In other words, you have learned to see it.
  4. Hunting for bugs. Kei Abe came up with the “bug hunt” as a means of making managers aware of common small problems that are easily overlooked. 10 to 20 managers get each a stack of 10 red tags and 20 minutes to attach them to frayed cables, broken gauges or switches, puddles of oil, lubricant on the floor, devices held in place by duct tape, or any other detail that is clearly wrong. Wherever I have seen this method used, all managers used up their stack of tags, and came away stunned by the sheer number of small maintenance problems they found.
  5. Conducting video time studies. These directly generates process time data, and indirectly causes you to notice details of the work that you would otherwise miss. For example, you see that an operator is much busier than a neighbor, or lit from behind behind and working in his or her own shadow, etc.

Other perspectives on this topic include the following:

  • Eugene R. Goodson, in the May, 2002 issue of the Harvard Business Review, published an paper entitled Read a Plant — Fast proposing a method for a small team to rate a plant in 11 categories–including safety, scheduling, inventory, teamwork, and supply chain. I find the recommendations useful but insufficient, particularly in the areas of layout and work station design. I also don’t believe in assigning scores on subjective scales, with categories ranging from “poor” to “best-in-class.”
  • Joseph Paris has a blog post called The Gemba Beyond the Window, with some interesting insights on communicating the monetary value of what you have on the shop floor.
  • The AME’s Accelerating the Journey blog contains the following list of  “10 Questions asked on a Gemba Walk”:
  1. What are the business issues with this product?
  2. Who is responsible for the value stream for this product?
  3. How are orders from the customer received?
  4. Where is the pacemaker process, triggered by customer orders?
  5. How capable, available, adequate, and waste-free are assembly activities?
  6. How capable, available, adequate, and waste-free are the fabrication activities feeding assembly?
  7. How are orders transmitted up the value stream from the pacemaker process?
  8. How are materials supplied to the assembly and fabrication processes?
  9. How are materials obtained from upstream suppliers?
  10. How are employees trained in Lean procedures motivated to apply them?

I find these questions puzzling, for the following reasons:

  1. You don’t need to be on the shop floor to find the answers to questions 1, 2, 3, 4, 7 and 10.
  2. Questions 5 and  6, that are really about the shop floor, are leading. Asked in this form, they suggest that assembly and fabrication are indeed waste-free. What you are really after is finding out where the waste is, and where the processes lack capability or capacity.
  3. Question 4 implies that production is scheduled by heijunka on a pacemaker process and pull on other processes. The more general question is of how production is scheduled in the plant. As discussed above, only part of this answer is found on the shop floor.
  4. Question 10 implies that there should be a Lean training program. I don’t understand why training should be the only aspect of the company’s Lean program to rate a question. Before coming to the floor, I would ask whether the plant has such a program. On the shop floor, you should see its effects.

Safety Stocks: More about the formula

In a previous post on 2/12/2012, I warned against the blind use of formulas in setting safety stock levels. Since then, it has been the single most popular post in this blog, and commands as many page views today as when it first came out. Among the many comments, I noticed that several readers, when looking at the formula, were disturbed that three of the four parameters under the radical are squared and the other one isn’t, to the point that they assume it to be a mistake. I have even seen an attempt on Wikipedia to “correct that mistake.”

I was myself puzzled by it when I first saw the formula, but it’s no mistake.  The problem is that most references, including Wikipedia,  just provide the formula without any proof or even explanation. The authors just assume that the eyes of inventory managers would glaze over at the hint of any math. If you are willing to take my word that it is mathematically valid, you can skip the math. You don’t have to take my word for it, but then, to settle the discussion, there no alternative to digging into the math.

A side effect of working out the math behind a formula is that it makes you think harder about the assumptions behind it, and therefore its range of applicability, which we do after the proof. If you don’t need the proof, please skip to that section.

Math prerequisites

As math goes, it is not complicated. It only requires a basic understanding of expected value, variance, and standard deviation, as taught in an introductory course on probability.

In this context, those who have forgotten these concepts can think of them as follows:

  • The expected value E(X) of a random variable X can be viewed, in the broadest sense, as the average of the values it can take, weighted by the probability of each value. It is linear, meaning that, for any two random variables X and Y that have expected values,

E[X+Y] = E[X]+E[Y]

and, for any number a,

E[a\times X]= a \times E[X]

  • Its variance is the expected value of the square of the deviation of individual values of X from its expected value E(X):

Var(X) = E[X-E(X)]^{2}= E[X^{2} -E(X)^{2}]

Variances are additive, but only for uncorrelated variables X and Y that have variances. If

E[[X-E(X)] \times [Y-E(Y)]]= 0


Var(X+Y) = Var(X)+Var(Y)

  • Its standard deviation is

\sigma = \sqrt{Var[X]}

Proof of the Safety Stock Formula

Fasten your seat belts. Here we go:

As stated in the previous post, the formula is:


  • S is the safety stock you need.
  • C  is a coefficient set to guarantee that the probability of a stockout is small enough.
  • The other factor, under the radical sign, is the corresponding standard deviation.
  • μL and σL are the mean and standard deviations of the time between deliveries.
  • μD and σD are the mean and standard deviation rates for the demand.

  \sqrt{\mu{_{L}^{}}\times\sigma_{D}^{2}+\mu_{D}^{2} \times \sigma_{L}^{2}} is the standard deviation of the item quantity consumed between deliveries, considering that the time between deliveries varies.

μD and  \sigma_{D}^{2} are the mean and variance of the demand per unit time, so that the demand for a period of length T has a mean of \mu_{D} \times T , a variance of  \sigma_{D}^{2} \times T, and therefore a standard deviation of \sigma_{D} \times \sqrt{T}. See below a discussion of the implications of this assumption.

Note that the assumptions are only that these means and variances exist. At this stage, we don’t have to assume more, and particularly not that times between deliveries and demand follow a particular distribution.

If D(T) is the demand during an interval of duration T, since:

we have:

If we now allow T to vary, around mean μL with, standard deviation σL , we have:

and therefore:


That’s how the variance ends up linear in one parameter and quadratic in the other three!


 \sigma\left [ D \right ]= \sqrt{\mu _{L}\times\sigma_{D}^{2} + \mu _{D}^{2} \times \sigma_{L}^{2} }


Note that all of the above argument only requires the means and standard deviations to exist. There is no assumption to this point that the demand or the lead time follow a normal distribution. However, the calculation of the multiplier C used to calculate an upper bound for the demand in a period, is based on the assumption that the demand between deliveries is normally distributed.


The assumption that the variance of demand in a period of length T is \sigma_{D}^{2} \times T implies that it is additive, because if  T = T_{1} + T_{2}, then \sigma_{D}^{2} \times T = \sigma_{D}^{2} \times T_{1} + \sigma_{D}^{2} \times T_{2}.

But this is only true if the demands in periods T_{1} and T_{2} are uncorrelated. For a hot dog stand working during lunch time, this is reasonable: the demands in the intervals between 12:20 and 12:30, and between 12:30 and 12:40 are from different passers by, who make their lunch choices independently.

On the other hand, in a factory, if you make a product in white on day shift and in black on swing shift every day,  then the shift demand for white parts will not meet the assumptions. Within a day, it won’t be proportional to the length of the interval you are considering, and the variances won’t add up. Between days, the assumptions may apply.

More generally, the time periods you are considering must be long with respect to the detailed scheduling decisions you make. If you cycle through your products in a repeating sequence, you have an “Every-Part-Every” interval (EPEI), meaning, for example, that, if your EPEI is 1 week, you have one production run of every product every week.

In a warehouse, product-specific items don’t need replenishment lead times below the EPEI. If you are using an item once a week, you don’t need it delivered twice a day. You may instead receive it once a week, every other week, every three weeks, etc. And the weekly consumption will fluctuate with the size of the production run and with quality losses. Therefore, it is reasonable to assume that its variance will be \sigma_{D}^{2} \times T where T is a multiple of the EPEI, and it can be confirmed through historical data.

You can have replenishment lead times that are less than the EPEI for materials used in multiple products. For example, you could have daily deliveries of a resin used to make hundreds of different injection-molded parts with an EPEI of one week. In this case, the model may be applied to shorter lead times, subject of course to validation from historical data.