What Are We Learning Since We Started Learning to See?
By Mike Rother
The Learning to See workbook, which was first made public at the Lean Enterprise Institute's June 1998 Lean Summit in Hartford, CT, has surprised us all with its ongoing success. Learning to See has now sold over 85,000 copies in English, been translated into nine languages (most recently Spanish), won the 1999 Shingo Prize for research in manufacturing, and continues to garner acclaim as a guide for lean transformations.
At first glance, Learning to See is about a method and tool for analyzing and designing value streams, but our primary intent in writing the book was to help readers widen their perspectives from a limited focus on process-level improvement to include a view of the overall flow, or "value stream.” We designed the book's format with that objective in mind, and hope that it has indeed proved to be a perspective expander for you and your team.
The flow or value stream perspective represents a shift from vertical to horizontal thinking. Horizontal thinking means looking across the traditional vertical structures of functions and departments to connect activities in the stream of value flowing from suppliers, through the organization, and on to customers. In other words, concentrating on overall flow means focusing on system efficiency rather than on just the "point efficiency" of individual elements in your organization.
Building on Toyota’s Tool
In the last few years, a flow or value stream oriented manufacturing measurable has started climbing to the top of the heap: production lead time (or it's inversely-related sibling, inventory turns). One of the main goals of most production systems today is a continual reduction of lead time, which requires that processing steps in the value stream become more closely coupled to one another, allowing value to efficiently flow across them. Toyota's original flow mapping methodology — which we expanded into Value Stream Mapping — has provided us with an especially practical tool for thinking about flow and designing value streams with shorter lead times.
With all the interest and activity around value stream mapping it seems a good time to review how the value stream perspective and the mapping tool are developing, and some lessons we have learned along the way. That is the purpose of this article. I'd like to comment on some current issues, lessons learned and reader feedback relative to value stream mapping and value stream thinking.
I. New Developments and Issues in Value Stream Mapping
• The Paper Flow is Getting More Attention
When we say lead time, we are usually thinking about production lead time, which is the time it takes to go from raw material to shipment, or from "dock to dock.” However, another type of lead time measurement is the order lead time, or the time it takes to go from a customer order to delivery.
Figure 1: Production and Order Lead Times
(Note: in make-to-order value streams, the schedule point will be further upstream than is shown here.)
When you take a close look at the flow of a customer order through many organizations, it doesn't take long to see that a significant amount of the order lead time often occurs in administrative functions; before the order even reaches the shop floor. This is another value stream, which is being called the "administrative" or "service" or "white-collar" value stream. Here too, material (i.e. orders) sits in inventory (i.e. in-baskets) and doesn't flow. As long as an administrative process is truly necessary, then we should also apply horizontal thinking and perspective here in order to design administrative value streams with short lead times.
The value stream mapping tool presented in Learning to See is increasingly being applied to administrative flows. Of course, it often takes longer to personally trace a practically invisible administrative value stream, but that's how you have to do it, because you should not rely on statements like "this is how it normally goes.” Administrative processes are also usually more make-to-order in nature, and thus different from high-volume production processes. But these obstacles should not prevent you from getting started with flow analysis and improvement across administrative processes.
In addition to the order flow, there are, of course, several other administrative and support activities in an organization, such as processing engineering changes, maintenance, quality assurance, and so on. All of these can be analyzed and redesigned with the goal of shortening their lead times. However, in such other areas, which exist primarily to support or enable the production flow, there are two additional questions that you need to be asking:
Is this activity necessary?
How should this activity be conducted in order to support the shortest possible production lead time?
In other words, never forget that the value-adding production flow is the customer that these support functions should be serving.
• Readers are Expanding Their Value Stream View into Supply Chains
Once you have initiated flow and lead-time improvement inside the four walls of your facility, you can start expanding your view to include the supply chain. More and more users of value stream mapping are reporting successes in this area, and LEI's new workbook, Seeing the Whole by Dan Jones and Jim Womack, gives you more insight and suggestions on what some are calling "macro mapping."
At one level, value stream refers to the entire flow from raw materials coming out of the earth all the way through to the hands of the end consumer; and increasingly even beyond that to include recycling the product. However, if this is too much to tackle right now then I suggest you expand your one-facility view by going from the point of use at your customer's facility back through to the receiving dock at one or two of your most important suppliers. The discoveries of waste, batched information flow and interrupted material flow that you have made inside your own facility will repeat themselves there, and issues of location and transportation between facilities will become additional factors.
Although a single-piece flow across the supply chain is usually still a dream, closer-coupling concepts like pull systems between facilities and milk-run deliveries can be applied with dramatic results. Even at the supply-chain level the basic lean goal remains the same: How can we get ever nearer to having each process make (and, if necessary, the delivery truck pick up) only what the next process needs when it needs it?
• The "Pacemaker Process" Needs More Attention
A question that often arises is, "What should we focus on as we analyze and redesign our value streams?" There are a host of factors that affect lead time (take a look at pages 30-36 in Richard Schonberger's new book, Let's Fix It), but clearly the pacemaker process — which is often a final assembly process — is one that needs more of our attention. Many manufacturers don't recognize the pacemaker process' important role in attaining a short lead time through the value stream.
Today's in-plant value streams can often be divided into two segments: pacemaker and fabrication. Definitions:
The upstream fabrication processes respond to requirements from internal customers, and often utilize general-purpose or shared equipment to produce a variety of components for different downstream processes.
In contrast, the downstream portion of a value stream is often dedicated to a particular product family and responds to external customers. This segment typically starts at the value-adding process that is the schedule and leveling point in a lean value stream (see Learning to See p.86), and involves processing steps that give the product its final form for the customer. This downstream segment of the value stream is called the pacemaker.
Figure 2: Fabrication and Pacemaker Segments of a Value Stream
The pacemaker process influences production lead time because it is the rhythm-setting point, or "heartbeat," for the value stream. If the pacemaker makes large batches of one product type, or if it has significant fluctuations in production volume, then the upstream fabrication processes will have to hold more inventory to be able to meet the peaks of the pacemaker's jerky component requirements.
In addition, due to the "bullwhip" effect (which was first described by Jay W. Forrester in 1958) any mix or volume surges at the pacemaker get amplified as you move up the value stream and into the supply chain. The effects of pacemaker fluctuations get worse the further upstream you go! (The Seeing the Whole workbook has more information on how you can assess and reduce the effects of such demand amplification in your supply chains.)
Another problem: when the pacemaker makes large batches of one product type, then your external customers who want other types have to wait, or you will have to try to hold even more finished goods inventory of items that you think customers will want. (And correctly guessing what customers will want in the future is difficult to do.)
All of this means that the efficiency of your value stream depends partly on how small you can keep the volume fluctuations and batches at the pacemaker (assembly) process. Many plants are trying to link their chain of processes by establishing supermarket pull systems between processes. However, if you run significant volume fluctuations and/or large batches in assembly, then the inventory in those supermarkets will be too high. Result: with or without pull systems the lead time through the value stream will still be long. Leveled and mixed production at the pacemaker — a steady heartbeat — helps make shorter production lead time possible.
Think of the pacemaker process as the conductor of an orchestra. To achieve lean value streams, managers, production control, maintenance, supervisors and engineers will need to pay closer attention to how you are operating your pacemakers. (Please refer to LEI's Creating Continuous Flow workbook for detailed guidelines on setting up and running operator-based pacemaker process.)
As you may have guessed, the final assembly point in a macro value stream (across several facilities and companies) is the rhythm-setting pacemaker process to which the supply chain responds. The characteristics of the information flow emanating from this point will influence a how lean a whole upstream supply chain can be.
• Leaner Value Streams Will Require Faster Response to Problems
As the lead time through a value stream shrinks, the processes in that value stream become more "close coupled" (less buffer between them). This makes a value stream more sensitive to problems. When there is a machine breakdown, absenteeism, defective parts and so on in one segment of a lean value stream it will take less time before these problems adversely affect other segments. This is especially true of problems at the pacemaker process (see above).
As it more closely couples its value streams, industry will need to work more on a heretofore largely ignored aspect of Toyota-style manufacturing: The leaner a production chain gets, the greater the need for swift, local response to abnormalities. And swift response requires swift awareness of abnormalities, someone to do the responding and a structured approach for how to respond.
A critical issue here is how management thinks about production problems. If our vision is to ban problems from production (indicated by statements like, “We just need more discipline”), then we will organize and manage differently than if we assume that problems are going to occur. There seems to be a tacit belief that Toyota’s system means the occurrence of problems will be eliminated. Yet Toyota and its suppliers also experience quality problems, machine breakdowns, absenteeism, and so on. In fact, these types of problems are statistically guaranteed to occur. And as one problem is solved and its causes eliminated, others will develop. New product programs also bring new problems.
When management assumes that problems in production are inevitable, then the most important question becomes, “How will we respond to them?” This is going to become a particularly important question for any company that is seriously trying to adopt lean production, since the faster you can successfully respond to abnormalities, the leaner you can make your value stream.
II. Notes and Advice From the Field
Learning to See has been very successful, but at the same time any book is subject to various interpretations by its readers. Here are some that we have observed as we visit and work with many companies.
• Some readers of Learning to See appear to think that value stream mapping is in itself a goal, occasionally telling us, "We are drawing maps of all our value streams!" That may lead to a better understanding of your flows, but not necessarily to any measurable improvement. Think of mapping as a method and tool to assist you in making flow improvements. The more important goal is active implementation of an improved future state.
Instead of mapping everything and expecting good things to happen as a result, we suggest you, 1) select a value stream where business objectives require measurable improvement, 2) develop an understanding of the current state, 3) design and agree upon an improved future state that can be introduced within six or 12 months, 4) put together an action plan to implement that vision (who is responsible for achieving what elements of the vision with what measurable results by when?). And in 12 months, you need to create the next future state map for that value stream and again charter a team to achieve that future state.
• We would also like to mention that having a perfectly drawn current state map is not the initial focus of VSM. A main intention of current state mapping is the process of understanding the dock-to-dock flow — the looking and sketching and trying to see and comprehend what is happening — and thinking about what should be happening — as material and information travel through your facility (or supply chain). Have you noticed that the person who best understands the material and information flows that a value stream map represents is the person who drew the map? The map creation process —more than just the map itself — helps you learn to see.
This is why we suggest that you begin sketching with just pencil and paper as you walk a flow. The looking, sketching and resketching may seem like a lot of manual work, but it is in fact a PDCA (Plan, Do, Check, Act) learning cycle that keeps you focused on the flow and deepens your understanding of the production system. With practice, you may be surprised at how quickly you can pretty accurately sketch dock-to-dock material and information flows. In other words, how quickly you can see. Then you are in a better position to develop useful future state value stream designs.
• There has occasionally been a mix-up between value stream mapping and traditional process flow charting, which is typically used by industrial engineers to analyze and improve a process. Value stream mapping (and design) cuts across process, departmental and functional boundaries, as well as across existing departmental performance-measurement systems. It involves trying to optimize dock-to dock flow instead of an individual process.
Since the value stream cuts across boundaries it means that top manufacturing management should be leading the value stream improvement effort. Although they may not necessarily do the mapping themselves, manufacturing executives and managers should be able to read a future-state value stream map and know what questions to ask.
• Some readers have focused intently on counting and tallying inventory, and using that data to estimate the production lead time as described in Learning to See. Lead time is a great metric and we recommend that you focus on reducing it. (Note: Outsourcing lead time does not equal reducing it.) However, don't let the activity of counting inventory become more important than the fundamentals of seeing and understanding the flow (or lack of flow) through your value streams.
Inventory tells you where a flow is interrupted. Once you have found these spots in your value stream then the next questions are, "Why does the flow stop here?" (there is always a reason), and, "What can we do to improve this situation?"
• Sometimes readers find it difficult to stay at the 100 ft.-level altitude the first few times they map a value stream to understand the current state. After years of making process-level improvements, you may naturally tend to drop to a very detailed level of analysis at every process along the way; trying to record all sorts of current-state process data up front. This especially seems to happen when you walk the flow through your own, familiar facility. Unfortunately, you then tend to lose the valuable overall flow perspective that Learning to See is all about.
My suggestion? Begin with the main or most important dock-to-dock branches of your value stream, and walk through them at a "100 ft. altitude.” Then go back and progressively drop down to add detail or additional branches on successive walk-throughs as is necessary to support your design and implementation of a future state. The first walk-through may only take an hour and result in only a rough current-state sketch.
The mapping process is not linear, although our presentation of it in Learning to See might lead you to believe that. You don’t really finish a current state map (“done!”), then finish a future state map (“done!”) and then shift to implementation. There is considerable overlap and feedback between these stages, and you should expect to have to periodically go back and gather more data as you realize you need it.
• We've noticed some readers going too far out into the future with their initial future state designs. Such future states are nice in theory, but about as difficult to achieve as a home run. Introducing Toyota-style production can't be done overnight, and involves steady progress via lots of base hits.
If you have drawn more than, say, six kaizen lightning bursts on a future state map, you are probably reaching too far out into the future with that map.
A suggestion is to draw one sketch of where you would like to be in about 5 years — we call this an "ideal state map" — and another, more detailed "future state map" of what you and your team can agree to have implemented within a maximum of 12 months time. Then as you work on implementing the 12-month future state you should be learning things and can fine tune your "ideal state" vision. At the end of 12 months it is time to have another 12-month map and implementation plan.
III. Reader Feedback
Our publisher, the Lean Enterprise Institute, has been open to comments and feedback from its community of readers. In addition to much positive feedback we have also received ideas and suggestions for improvements and additions to the Learning to See book, and were able to incorporate several of them in the current edition of Learning to See. There are also a handful of items that we have not added, such as:
More end-item variety to the Acme Stamping case.
Examples from different industries.
VSM in non-production settings. ("Administrative" or "white-collar" mapping.)
Detailed guidelines for managing the implementation of a future state value stream vision.
Tips for mapping across multiple facilities and organizations. ("Supply chain" or "macro" mapping.)
How to set up and run the pacemaker process.
Dealing with changing demand.
Calculating the financial benefits of a future-state design.
These — and probably more — could all be useful additions to Learning to See. But then by far the most frequent feedback of all is that the book is wonderfully straightforward, clear, and easy to understand and use. Readers are able to absorb the book and get started with positive action in their own facilities, and in many cases train others (sometimes using LEI's Training to See kit) in value stream mapping.
This makes us hesitant to complicate Learning to See with a string of detailed appendices in an attempt to make the book completely authoritative. Like any book, Learning to See reflects a need and some thinking at a point in time. The intent is to provide a springboard for seeing and thinking about your own value streams. Learning to See says what it says, does what it does, and we are pretty happy with that. You must be too, because demand for the book is still growing.
In fact, the amount of positive action, experimentation and dialog that is being triggered in part by Learning to See has been terrific. Many people are working on all manner of problems related to removing waste from their value streams. Follow-on books are being published at the Lean Enterprise Institute (such as Creating Continuous Flow, Seeing the Whole, Learning to Count), workshops are focusing on "value stream" and "value stream management," and elements of the simple (but surprisingly effective) Acme Stamping case from Learning to See keep popping up everywhere.
So Learning to See continues to serve as the fundamental value stream book, while supplements to it are being created in an even better fashion than just a couple of authors might do. Many more people are working on the issues, via several channels and with interesting perspectives.
Of course, successful implementation is what really counts. So, our hats go off to all who are rolling up their sleeves and advancing the war on waste through implementation that generates shorter lead times and positive results for customers. Keep on looking and thinking — and seeing — with a value stream perspective.
With best regards,
Mike Rother is the co-author of LEI's Learning to See and Creating Continuous Flow books, as well as the Training to See training kit. He is a teacher at the University of Michigan and a manufacturing consultant based in Ann Arbor, Michigan. Mike recently returned to Ann Arbor after spending a year as guest researcher at the Fraunhofer Institute for Production Technology in Stuttgart, Germany.
Copyright © March 2002, by Mike Rother
Label: What Are We Learning Since We Started Learning to See