In one sentence: The sequel to The Machine That Changed the World distills Toyota’s system into five principles (value, the value stream, flow, pull, perfection) and proves them with before-and-after numbers from real factories, where the same people and machines doubled productivity and cut lead times by 90 percent once the batches went away.
At a Glance
Author: James P. Womack and Daniel T. Jones
First published: 1996 (Simon & Schuster), revised and updated Free Press edition 2003
Category: Business / Operations & Management
Length: about 396 pages, about 125,000 words (Free Press revised edition)
ISBN-13: 978-0-7432-4927-0 (Free Press revised and updated edition, 2003)
Summary reading time: about 12 minutes
Book reading time: about 8 hours
Notable adaptations: none, though the book seeded the Lean Enterprise Institute, the Learning to See value-stream mapping workbook, and the lean movement’s spread into healthcare, software, and startups
Womack and Jones spent the first half of the 1990s inside more than fifty companies, after their earlier benchmarking study named Toyota’s method “lean production” and left readers asking how to actually do it. This book is the answer. Part I lays out the five principles. Part II walks through four full conversions, from a Kentucky wrapping-machine maker to Pratt & Whitney’s jet-engine empire and Porsche’s craft culture. The revised edition adds a 2003 epilogue tracking how those companies fared and what the authors got wrong the first time.
Read it if you run operations, manage any process with handoffs and queues, or keep hearing “lean” used to mean cost-cutting and want the original meaning back. The case studies assume manufacturing, but the authors apply the same logic to air travel, healthcare, groceries, and construction. Skip it if you want a quick airport read. This is a dense, numbers-heavy argument that rewards patience and punishes skimming.
The Big Idea
Muda is Japanese for waste: any activity that absorbs resources and creates no value. The book’s claim is that most of what organizations do is muda, not because managers are incompetent but because the batch-and-queue logic they manage by is wrong. The cure is five linked principles. Specify value from the customer’s point of view, not your own. Identify the whole value stream for each product and strip out the steps that create nothing. Make the remaining steps flow continuously instead of waiting in batches. Let the customer pull work through the system rather than pushing forecasts into it. Then chase perfection, because each round of improvement exposes waste the last round hid. Applied together, the authors’ rule of thumb is that converting a classic batch operation doubles labor productivity, cuts throughput time and inventories by 90 percent, and halves errors, with very little capital spent.
Key Ideas
1. Only the customer gets to define value
Companies define value by what they already make, engineers define it by what impresses other engineers, and whole countries have their own distortions. The opening examples make it concrete. One author’s 350-mile trip took seven hours door to door, two of them taxiing at a hub and shuffling through terminals, because airlines optimize airplanes and airports rather than the passenger’s journey. The Jones family’s package holiday to Crete involved 23 processing steps, ten lines, and eight inspections asking the same questions, since each of 19 organizations optimized its own fragment. Texas builder Doyle Wilson won a national quality award, then discovered 78 percent of local buyers chose older homes just to avoid the hassle of dealing with builders like him. He’d been competing brilliantly for a fifth of the market. Specifying value means asking what the whole product is worth through the customer’s eyes, then setting a target cost based on what the product would cost with the visible waste removed, and using the gap as the agenda.
2. Map the value stream and brace yourself
Follow one product through every step from raw material to customer and three categories appear: steps that create value, steps that create none but can’t yet be avoided (Type One muda), and steps that can go now (Type Two muda). The book’s famous specimen is a carton of cola. Bauxite mined in Australia becomes a can on an English shelf through smelters, rolling mills, can makers, and warehouses: 319 days of elapsed time containing about three hours of actual value creation, with the material picked up and put down 30 times. More than 99 percent of the time, the value stream isn’t flowing at all. Nothing in that chain is run by fools. The problem, the authors insist, is the logic itself, giant fast machines that are efficient locally and ruinous for the whole. The 2003 edition’s windshield-wiper map repeats the lesson across four companies: 8 of 73 physical steps created value, none of the 25 information steps did, and demand that varied 3 percent at the customer swung 40 percent at the raw-material end.
3. Make value flow, and abolish the department to do it
Batches feel efficient and aren’t, which the authors illustrate with a newsletter-folding experiment where children insist on folding all, then labeling all, then stamping all, “because that wouldn’t be efficient” otherwise. Most of the world runs on the same instinct. Flow means taking the specific product, never letting it out of sight, and ignoring the boundaries of jobs, departments, and firms so it moves without stopping. Henry Ford got there first, cutting Model T assembly effort 90 percent in 1913, but his flow needed identical products at huge volume. Toyota’s Taiichi Ohno and Shigeo Shingo extended flow to small lots through quick changeovers and right-sized machines. The toolkit that follows is now standard vocabulary: takt time to pace production to the rate of demand, U-shaped cells of different machines in tight sequence, standard work written by the team that does it, visual controls so status is obvious at a glance, and mistake-proofing so defects can’t travel. One warning recurs: if you think you need to spend big money to convert to flow, you don’t yet understand lean thinking. Pratt & Whitney replaced an $80 million automated grinding monument with a cell of simple machines that cut throughput from ten days to 75 minutes.
4. Pull means making nothing until someone asks
Flow without pull just makes unwanted goods faster. The pull chapter traces a single bent Toyota pickup bumper from a Pennsylvania dealer back through the chain. Before lean, the system held nearly eleven months of bumpers in warehouses and lots. After supplier Bumper Works cut press changeovers from 16 hours toward 10 minutes and linked each stage with kanban signals, steel arrived and shipped as a finished bumper within 48 hours, and the whole chain ran on “sell one, buy one” replenishment loops. Toyota’s parts warehouses doubled picker productivity against Big Three rivals, and a dealer who adopted the system halved his parts inventory while adding four service bays in the freed space. The authors push the logic further: much of what looks like market chaos is self-induced, created by long lead times, deep inventories, and promotions that make demand lurch. Cut the lead times and level the selling, and the lurching calms down.
5. The transformation playbook: a change agent, a sensei, and a crisis
Part II is the book’s proof, and the numbers repay quoting. Lantech, a Kentucky stretch-wrapper maker bleeding from a lost patent suit, replaced its departments with four product cells: throughput fell from 16 weeks to 14 hours, defects per machine from 8 to 0.8, on-time shipment rose from 20 to 90 percent, and share grew from 38 to 50 percent with essentially zero investment. At Wiremold, CEO Art Byrne personally taught the courses and ran kaizen events: inventory turns went from 3.4 to 15, sales per employee doubled, and $11 million of freed cash bought five companies. Pratt & Whitney, hemorrhaging after the Cold War ended, cut unit costs 20 percent while volume halved and swung from a $283 million loss to a $530 million profit. Porsche, whose craftsmen took pride in rectifying cars for days at the end of the line, halved inventories within two years and in July 1994 rolled out its first defect-free car in 44 years. The pattern across all four: a change agent with real authority, a sensei (often Ohno’s own students) setting targets that sound impossible, a crisis that unfreezes the organization, results demanded in the first week, and the same activity improved again and again. The authors’ timetable is sobering: two to three years for the rudiments, five to institutionalize.
6. Face the jobs question once, honestly
The book is unusually candid that lean frees people, typically three quarters of the effort in a batch conversion, and that pretending otherwise kills the effort. The working rule from the cases: size the workforce to reality once, up front, then guarantee that nobody loses their job from improvement activities, and grow the business to absorb the freed capacity. Wiremold cut 30 percent through early retirement and delayering, then guaranteed the rest and doubled sales. Porsche negotiated one reduction of 2,500, then extended a jobs pledge. Pratt, with no guarantee, went from 51,000 to 29,000 people. What never works is “drip torture,” improvement followed by layoffs followed by improvement, which teaches everyone that cooperating means firing yourself. The authors also map the failure mode for middle managers, the roughly one in ten “anchor-draggers” who can’t accept a system that is, in Byrne’s words, profoundly corrosive of hierarchy.
7. Perfection is a direction, not a destination, and the 2003 epilogue shows the compounding
The fifth principle sounds mystical and operates practically: keep a picture of perfection in view, compete against it rather than against competitors, and pick a few policy-deployment goals a year instead of thirty. Freudenberg-NOK ran six kaizen events on one line over three years and gained 991 percent in productivity, prompting a skeptic to ask why they didn’t get it right the first time, which is exactly the misunderstanding. The revised edition’s epilogue tracks the compounding. Wiremold, worth $30 million near bankruptcy in 1991, sold for $770 million in 2000. Porsche hit a 17 percent return on sales in 2002, twice Toyota’s. Toyota itself marched from 10 to 11 percent global share against a target of 15. The authors also correct themselves: their original value-stream maps were too simple, the value-stream manager role mattered more than any reorganization, and process-level improvement without a system-level picture produces what they christened “kamikaze kaizen,” lots of commotion and no benefit that reaches the customer. Their “lean math” closes the epilogue, an argument that offshoring calculated on piece price alone ignores inventory in transit, remainders, expediting, warranty visits, and currency risk, so production usually belongs near the customer.
Context and Analysis
Lean Thinking sits beside a small shelf of books that changed how operations are actually run, and its vocabulary (value stream, muda, pull, kaizen event) became the industry’s vocabulary. It pairs naturally with Goldratt’s The Goal, which dramatizes the same war on batch logic through the theory of constraints, and its insistence on disciplined, evidence-first transformation over charisma echoes the findings in Good to Great. The book’s intellectual move, elevating Toyota’s shop-floor techniques into five universal principles, is what let lean jump from car factories into hospitals, software teams, and startups.
The criticisms are real and worth weighing. The cases are hand-picked winners, and the authors say the methods “always work,” a claim their own sample can’t carry. Several headline numbers are projections made as the first edition went to press, and confounders run through the stories: Lantech’s doubling rode a market recovery, Pratt’s profit rebound came with 22,000 fewer employees and outsourced work, which relocates cost as much as it eliminates it. The no-layoffs principle, on inspection, holds only after heavy initial cuts. Labor scholars have long argued that takt-paced work with thin buffers intensifies pressure on workers, and the book’s answer, that flow work is psychologically satisfying, is the authors’ own assertion, with little independent worker testimony behind it. Some history is smoothed (the supermarket origin story for just-in-time is disputed, and statistics like half of US books being shredded arrive without sources). The authors do disclose an ownership stake in the bicycle company they use as an example, and the 2003 epilogue’s self-corrections are to their credit. None of this has dislodged the core finding, which has replicated across thousands of implementations: batch-and-queue hides enormous waste, and the five principles find it.
How to Apply It
Pick one product family and map its value stream this month, on one page, with a stopwatch and a walking tour rather than software. Record every step, its time, and the waiting between steps, then sort the steps into value-creating, unavoidable for now, and removable immediately. Expect the ratio to embarrass you. Draw the information flow on the same map as the product flow, the connection the authors admit they originally missed.
Compute a takt time for the work by dividing available time by the rate of demand, and compare it to how the work actually arrives, in lumps and rushes. Then run one focused improvement event: take a single sequence of steps that sit in different departments, put them next to each other for one week, and process work one piece at a time. Demand a visible result within the week. If nothing improves, the diagnosis is wrong or the commitment isn’t real.
Before scaling up, settle the people question in writing. Decide what happens to anyone whose work the improvement eliminates, and make redeployment a promotion into the improvement team, not a waiting room. Choose a few goals for the year, three or four, and visibly deselect the rest. And when a step is fixed, schedule the date you’ll fix it again, because the first pass only lowers the water enough to show the next rocks.
Memorable Lines
“Muda. It’s the one word of Japanese you really must know.” (James P. Womack & Daniel T. Jones)
“In short, lean thinking is lean because it provides a way to do more and more with less and less—less human effort, less equipment, less time, and less space—while coming closer and closer to providing customers with exactly what they want.” (James P. Womack & Daniel T. Jones)
“Our earnest advice to lean firms today is simple: To hell with your competitors; compete against perfection by identifying all activities that are muda and eliminating them.” (James P. Womack & Daniel T. Jones)
“Where’s the factory? This is the warehouse.” (Chihiro Nakao, on first seeing Porsche’s plant, quoted by the authors)
“Perfection is like infinity. Trying to envision it (and to get there) is actually impossible, but the effort to do so provides inspiration and direction essential to making progress along the path.” (James P. Womack & Daniel T. Jones)
“This strategy has worked and continues to work because Toyota is the most brilliant manager of core processes in the history of industry.” (James P. Womack & Daniel T. Jones)
Should You Read the Full Book?
Verdict: Recommended
Anyone who manages a process should read this summary’s five principles twice, but whether to read all 396 pages depends on what you’ll do with them. If you run operations, a plant, a clinic, a kitchen, or any business where work waits in queues, the full book approaches essential, because its power lives in the case-study detail: the exact sequence Lantech followed, what Byrne said to a frightened workforce, how Pratt handled the managers who wouldn’t move, the specific failure modes in the 2003 epilogue. That texture is what makes the ideas executable rather than inspirational, and no summary can carry it all. General readers who want the worldview, that most activity is waste and the customer defines value, will get it here and can stop, skimming past Part II’s blizzard of percentages without guilt. The Lean Thinking book page has the full details and where to get a copy.