In part two of his article, Richard Schonberger gives LMJ readers an account of the history of lean.
The era of JIT manufacturing in Western countries was fairly easy to characterise in part 1 of this article. Its methodologies had evolved hardly at all from the beginning, circa 1981, to 1990 when JIT got renamed lean manufacturing. Though the lean era began with the same concepts and methods, over time lean has been amended and appended, with changing priorities and emphases. That shifting is substantial.
In part 1 I mentioned strong inventory-based evidence of lean’s up and down performance since the 1980s. I’ll provide that evidence, specifically that JIT had lost its way in the late 1980s and early 1990s; got rejuvenated in the later 1990s and through the 2000s under the lean banner; and then lost ground again in the most recent decade. What has been going wrong and right in lean manufacturing? My impression is that many lean success stories of recent times have been in plants that had similar lean successes years earlier, lost them, and then had to re-learn and re-apply lean all over again. This rework takes place under a mostly new management team, the old team being gone, taking their what works, what doesn’t experience with them.
To trace lean’s long-term zig-zag performance, I rely on a widely-used lean indicator: inventory. It serves well – for inventory-intensive companies – because it is a standardised, hard-data metric, and, for publicly traded companies, audited inventory records are publicly available. My own “leanness studies” began with plotting of inventory-turnover trends for a few dozen manufacturers in the US, UK and France in the mid-1990s, with data reaching back at least 15 years.
That database has grown to around 1,600 manufacturers, retailers, and distributors from many countries around the globe. Rather than absolute values, the research measures long-term trends. The intent is to show ingrained capability in which an improving trend is sustained even as products, executives, and competitive and economic conditions change. Graphs of inventory turns for all the companies are inspected and scored, each for a time span of at least 10 years; graphs and scoring are updated with fresh data every year. For example, a clear upward trend in inventory turns scores 2 points; and for a downward 10-year trend, minus ½ point.
Figure 1 extracts average long-term scores for just US and Japanese companies: for those two countries the grand database includes far more companies dating back far more years than for other countries. A larger sample, of course, yields higher-validity findings.
Oldest data come from 176 US companies dating from the 1950s (some are no longer in the database for reasons such as bankruptcy, merger, or dissolution). Average scores for those venerable companies revealed a down trend from the 1960s until the mid-1970s. That worsening may be owed largely to complacency: globally dominant, post-WWII US industry became lazy, leading to growth of bureaucracy, organisational silos and management by remote control, as applied to quality, maintenance, suppliers, costs, everything.
Getting humbled in the 1970s by the Japanese Juggernaut triggered the awakening, first in the quality arena, then as JIT production kicked in. The result, gleaned from scores for a database grown to a few hundred manufacturers, was sharply improving longterm inventory turnover scores. Later that trend ended and the numbers turned downward, suggesting JIT fatigue: executives, always looking for what’s next, lost enthusiasm for JIT; consultants, always ready to supply what’s next, were complicit. Meanwhile, total quality control, morphed into total quality management, had been watered down to where TQM was becoming the property of the organisationalbehavior community – more for teaming than quality improvement.
In 1990 lean arrived as a worthy what’s next. JIT, by the new name, was rejuvenated, and by about 1995 long-term inventory turns again were on the rise. It did not last. In the 2000s the scores again worsened, as, apparently, lean fatigue set in.
As Figure 1 shows, Japan’s had its own ups and downs. Sufficient companies and years’ data are lacking for the 1960s and into the 1970s. For that period, I’m assuming that Japan’s trend was upward. JIT had been developed there, and was well known and expanding beyond automotive. Moreover, manufacturers surely would be ready converts, given Japan’s congested environs and lack of natural resources: scarce, costly resources, husbanded and shrunk through JIT, prominently include materials, space, and transport, along with scrap and rework.
By the late 1970s the database held more than 100 Japanese manufacturers with data from at least 15 prior years. From then to the later 1980s the long-term scores improved, before worsening, then plateauing in the 1990s and early 2000s. This again suggests fatigue – JIT/TPS (rather than lean) fatigue since the term lean, not of Japanese origin, has not been widely popularised there. That fatigue may have been aggravated by occurring during Japan’s long period of economic stagnation: historically it has been socially difficult in Japan to reduce capacity. So when times got tough, manufacturers kept producing and growing inventories, with lean practices falling by the wayside. Finally, Japan’s inventory numbers have been improving, partially explained perhaps by “lean” implementation in the form of capacity reduction and off-shoring.
LEAN’S CONFUSED EVOLUTION
Evidence from Figure 1 indicates that lean works. And that it doesn’t. That raises questions of why. Categorising lean as to its primary and secondary elements and tendencies may be helpful in clarifying the issues.
First of all, the core of lean as it arose in 1990 consisted of the same seven (by my own reckoning) methodologies that were dominant in JIT era. The most potent three, originating largely at Toyota, are cells/focused factories, quick setup/small lots, and kanban, each separately, better yet together, capable of magnitude reductions in production throughput times and inventories. The other four – closely supporting and enabling, and of various origins – are cross training/job rotation, right-sized equipment, quality-at-the-source, and total productive maintenance.
These seven are singled out in part for their prominence in JIT and lean writings, but more so because of their strong competitive impacts: by reducing throughput times they get product into the hands of customers faster with quicker discovery of defects and causes. This is the competitive edge that George Stalk, Jr., was talking about in his classic 1988 article, “Time—The Next Source of Competitive Advantage” (Harvard Business Review).
Table 1 categorises the seven methodologies and aligns them with beneficial results. Short cycle time is the standout benefit of cells/focused factories, and right-sized equipment is a specific enabler. At Mercury Wire in Massachusetts conventional cable-making machines were replaced by right-sized equipment for a cell dedicated to a family of products for a single customer. They assert that much smaller, much slower equipment moves orders through the cell 99% faster – with one-day turn around – because machine speeds are matched, enabling one-piece flow and no WIP.
Flexible response comes primarily through small lots made economical through quick setup, and supported by a versatile, quick-change, cross-trained, job-rotating workforce. Queue and queue-time limitation is achieved by kanban, which also synchronises the flow and adjusts it to process constraints. Quality is gained especially through quality at the source, and also as a by-product of TPM. Interruption avoidance obtains through quality at the source (fewer stoppages for nonconformities) and TPM (fewer equipment stoppages).
In certain situations a supportive/enabling methodology could be deserving of an upgrade to primary. One example: right-sized equipment at Mercury Wire was key to delivering 99% shorter cycle times for one customer. A versatile, cross-trained workforce might be the primary way to achieve flexibly quick response in some labor-intensive operations. Quality at the source may outweigh all else in lean’s quick-response agenda in cases where defects would trigger a disastrous, long-running sequence of corrective actions (the Deepwater Horizon oil spill comes to mind). The same could be said of TPM in nuclear power plants.
Numerous additional methodologies associated with lean, but generally secondary as to competitive impacts, are of two types, applied (do it) and analytical (study it). Table 2 is a non-comprehensive listing. The applied are so numerous they take up two columns. I’ll limit my discussion to a few highlights and overviews.
The table is organised into five numbered segments in the applied category. 5S, under number 1, since it includes keeping things neat and clean, could be treated as an element of total productive maintenance. But 5S can be less – or more – than that. Less when not accompanied by other lean elements. More when it blends with kanban as marked-off, sometimes color-coded kanban zones on floors, benches, shelves, and trays. The linkage of 5S to TPM or kanban makes up a visual workplace competitively delivering quicker customer response; in contrast, passive features of visual workplace mostly and merely involve score keeping.
Making up the second segment are six items involving lean supply and distribution, which deserve far more lean effort than they get. As my leanness database clearly shows, in most industries much more inventory and customer lead time exists in logistics than in factories.
The third segment includes ABC/lean accounting and backflushing, replacements for conventional accounting practices that create incentives contrary to JIT/lean, including producing in excess of and out of synch with usage. DFMA is a powerful lean methodology by itself, since in standardising parts it shrinks an unlean landscape of many types of resources producing many kinds of parts and products.
The eight applied items in the fourth segment – fail-safing through standard work – require no discussion since they enjoy high familiarity in today’s lean community. The fifth segment contains the remaining 15 – underscheduling through reduce number of touches. These applications once were familiar to consultants and authors, but today appear to be seldom or irregularly written about or spoken of.
THINK OR DO
In the analytical category the 13 items, all admirable to be sure, are in a sense non-value-adding. By that I mean it would be better if there were no need to map, chart, simulate, or projectise. Instead just do it. In plentiful cases, of course, just do it would be just doing it badly.
It does seem, though, that there has been a trend away from doing and toward excessive analysis. The lean lists from my Google search, as well as lean books and articles of the past decade or so, suggest that people are tending not to distinguish between think and do. That is, the analytical tools and those that actually lean out are treated as of the same stripe.
My collection of more than 1,000 lean case-study articles, largely from US periodicals (including Target, Industry Week, Industrial Engineer, etc) and summarised in a computer file, often cite, as main accomplishments, the extensive use of the analytical tools. Moreover, most of the discussions are about the doings of the professional and technical staff, and little about engagement of the work force.
As for this apparent tendency of staff people to elevate their roles in lean, leaving shrunken or token involvement of line people, self-interest may be at work. For one thing, lean is fun. For another, lots of lean experience looks good on one’s resumé. The counter is that the cv might look even better if it indicated skill at harnessing the common sense and experience of the masses of lower-cost production people as the primary vehicle for lean achievements.
For just do it to thrive, it must engage an empowered, lean-savvy workforce. If the associates have been trained in the key methodologies of Table 1 and the applied of Table 2, they will never run out of action-zone implementations, many of which they can do on their own. Where’s the time for that? Under-scheduling, first item, fifth applied segment in Table 2, is an excellent way to provide it: schedule production for, say, 7 ½ hours per shift, leaving the last 30 minutes for making schedule on a bad day, and on a normal day leaving time for associates to sip tea and work on improvements together.
The majority of consultants and lean authors would agree, I think, that most early efforts in setup reduction, TPM, and 5S can and should be handled by production associates. No engineering degree is needed to design shadow boards, clean and lubricate equipment, and set up labeled places for everything. Similarly, many consultants would have operators implementing common-sense quick-setup actions before calling in engineers to modify tooling.
While learning the methodologies, especially those in the applied category, is necessary, it is not sufficient. Motivation avidly to apply lean training also requires learning why. That is, the training must present a strong case to the workforce that most of the listed lean methodologies yield results quickly noticed by customers, in the form of shortened lead times. It takes a little longer, but customers notice as well that problems – quality or otherwise – are found out and dealt with more quickly. If training provides only methodologies, the missing “why” subverts the lean effort.
BACKSLIDING AND SIDETRACKING
Returning to the apparent phenomenon of lean/TPS/JIT fatigue, what can be done to avert it and make lean endure and thrive?
The term functional drift emerged from an article by Hyer and Brown positing that cells have a life cycle. Though clearly and hugely beneficial competitively, cells may be countered by human tendencies to get lazy. New customers, product and component designs, and suppliers upset the existing lean order. Out of laziness or the press of time, a new order gets bounced around to equipment in scattered cells. With more of that, a cellular configuration slides back to functional.
We might add batch drift. Small-batch production, attained through quick setup and right-sized equipment in cells, all designed for one set of product order types and volumes, are upset when new ones arise. The effect is batch drift.
Besides drifting, lean has developed tendencies to get sidetracked. I refer to cases where lesser, maybe easier, lean methodologies get priority, and primary ones – cells, quick setup, small lots, kanban, and so forth – get left behind.
A case in point comes from an article in a previous issue of LMJ, by Daryl Powell. It cites a survey of lean practices in Norway, stating that though “many of the fundamental lean practices (visual management, 5S, value-stream analysis, A3 reporting) had been applied” . . . there was no evidence of the uses of “flow production concepts or pull production.”
That is not a surprise, because, in the absence of flow/pull methods, visual management, 5S, and value-stream analysis come up short as lean standard-bearers (as for A3, if it is to qualify as a lean methodology then so should storyboarding, videoconferencing, and any number of other respected ways to frame issues).
The article notes that Norway is dominated by low-volume, high-variety, as well as continuous-process manufacturing for which kanban-driven pull production is ill-suited. True enough. However, in most cases of low-volume, high variety production quick-setup/ small-lot production is highly beneficial. And in continuous-process manufacturing TPM is commonly the most important of the lean practices.
To my mind, the best ways to avoid backsliding and sidetracking are by systematising lean-driven continuous improvement and elevating lean to the level of permanent company strategy. Systematising means standard, repeating planning and control routines, driven by a standard, repeating training regime. When lean is standard operating procedure, it takes little effort to sustain itself. With repetition features – daily, weekly, monthly – lean becomes ingrained.
As a final observation on the lean era, I must express my dismay about lean’s status in the organisation and the management hierarchy. We’ve been told that attacking the seven wastes is lean’s essence. For high executives, that relegates lean to an operations pursuit, to which it then is delegated downward. Recognising, redefining and promoting lean’s customer-side attributes – delivery of flexibly quick response and fast attention to problems – raises lean from operational to competitive to strategic.
Lean’s competitive attributes are compelling enough that marketers should become strong advocates. As strategy, lean elbows its way into the executive suite and the boardroom. More than that – unlike fleeting strategies, such as a new-market venture in Indo-China or across-the-board price cuts to stimulate sales – customer-focused lean should have staying power enough to become permanent strategy.