Menu
Book a demo
Yokoten: Meaning, How Toyota Spreads Improvements, and How to Prove It Worked

Yokoten: Meaning, How Toyota Spreads Improvements, and How to Prove It Worked

What yokoten means, how Toyota spreads improvements across plants, why fixes stall between lines, and a proof-based method with a worked money example.
Yokoten: Meaning, How Toyota Spreads Improvements, and How to Prove It Worked

In short: Yokoten is the lean practice of taking an improvement or a lesson proven in one place, such as a line, a shift or a plant, and applying it wherever the same problem exists. The word is short for yokotenkai, roughly "horizontal deployment". It is how one team's kaizen becomes a plant's or a company's standard, instead of staying a local success.

  • Yokoten spreads the know-how and the reasons behind a fix, not just a copied procedure. Receiving teams go and see, then adapt.
  • Research on internal transfer finds the main barriers are knowledge-related, such as unclear cause and effect, not just unwillingness.
  • Prove it on each receiving line with before and after data. In the example below, one changeover fix is worth $22,800 to $34,200 a year per line, five times over.

Guide, updated 14 November 2026. Sources are listed at the end.

What does yokoten mean?

Yokoten (横展) is the short form of yokotenkai (横展開), a Japanese business term. The Shogakukan Digital Daijisen dictionary, on Kotobank, lists both as synonyms of suihei tenkai (水平展開) and defines it as applying the knowledge, lessons, techniques and equipment gained in one business or case to other businesses or cases. Yoko means sideways or horizontal; tenkai means unfolding or deployment.

Yokoten sits next to kaizen. Kaizen improves one process; yokoten makes sure the same improvement reaches every process with the same problem. If you are deciding between small steps and a step change, see kaizen vs kaikaku; yokoten applies to both, because either kind of result can be spread.

In lean manufacturing the term is tied to Toyota. Toyota Material Handling, part of the Toyota group, renders it roughly as best practice sharing and describes it as horizontal, peer-to-peer transfer of information and knowledge across an organisation. Horizontal is the key word: the improvement moves sideways between teams doing similar work, not down from head office as an instruction.

Track OEE and downtime in real time
Book a demo

How yokoten works

A plain description of the cycle has five steps:

  1. Improve and prove locally. A team solves a problem on its line and shows the result in data, not opinion.
  2. Capture the know-how and the why. Write the new standard, the root cause it addresses and the conditions it depends on: machine model, product, material, crew size.
  3. Find where the same problem exists. Look for the same loss on other lines, shifts and sites. This is the step that turns a local fix into yokoten.
  4. Go and see, then adapt. People from the receiving lines visit the source line, see the change working and adapt it to their conditions.
  5. Verify on each receiving line and update the standard. Measure the same loss before and after on each line. Feed back what changed in the adaptation, so the standard gets better.
Yokoten: spread it, then prove itFive steps, one proof per receiving line1Improve and prove locallybefore and after, in data2Capture the know-how and whyroot cause and conditions3Find the same loss elsewherelines, shifts, sites4Go and see, then adaptpeer to peer, with a trial5Verify on each linethen update the standardExample: one changeover fix40 to 25 min, 3 a day, 250 days, crew of 4187.5 hfreed per linea year$22.8kto $34.2kper linex5more lines:$114k-$171kOne kaizen, paid for once.Yokoten makes it pay six times.
A five-step yokoten cycle with a per-line proof step, and the example value of one changeover fix at the BLS manufacturing wage and the FLSA overtime rate. See the source.

Step 4 is the one most often skipped. A fix sent by email as a procedure is a copy; a fix that people have seen running and understood is a transfer. That fits the peer-to-peer description Toyota Material Handling gives: the receiving team does the work of understanding, not head office.

How Toyota spreads knowledge

Two peer-reviewed studies describe the system yokoten sits in.

  • Inside Toyota. Youngkyo Suh's study Knowledge Network of Toyota (Annals of Business Administrative Science, 2017) finds that knowledge is created on the plant floor, that the Operations Management Consulting Division spreads both standardised and not yet standardised knowledge, and that the Global Production Center's main job is standardisation. The study concludes that the network keeps a balance between diversity and standardisation of knowledge.
  • Across suppliers. Jeffrey Dyer and Kentaro Nobeoka's study Creating and Managing a High Performance Knowledge-Sharing Network: The Toyota Case (MIT working paper, later in the Strategic Management Journal) identifies six institutionalised knowledge-sharing routines between Toyota and its suppliers, and argues that this network helps explain why Toyota kept its productivity and quality lead after its production system principles had spread through the industry.

The lesson for a plant is that yokoten is not a newsletter. It needs places where knowledge is created (lines that improve), people whose job includes carrying it (a CI team or line leaders with time), and a way to turn it into standards.

Why improvements do not spread on their own

It is a familiar story: one line halves its changeover time, and two years later the line next to it still runs the old way. Research helps explain why.

Gabriel Szulanski's study of internal stickiness (Strategic Management Journal, 1996) analysed 271 observations of 122 best-practice transfers in eight companies. Against the common belief that motivation is the main problem, it found the major barriers were knowledge-related:

  • The recipient's lack of absorptive capacity: the receiving team does not have the skills or time to take the practice on.
  • Causal ambiguity: nobody can say exactly why the practice works, so nobody knows what can be changed when adapting it.
  • An arduous relationship between the source and the recipient.

Public bodies show the same pattern. In GAO-02-195 (January 2002), the US General Accounting Office surveyed 192 NASA program and project managers and received 115 usable responses. 58% said current processes did not let them retrieve the right lessons at the right time, 27% had not known about the agency's lessons learned system before the survey, and 53% found lessons useful less than 25% of the time. The cultural barriers cited included a perception that lessons bring no benefit, lack of time, lack of trust and intolerance of mistakes.

Each finding points to a design rule for yokoten:

BarrierSourceYokoten design rule
Causal ambiguitySzulanski 1996Deploy the root cause and the reason, not only the new procedure. Name the conditions it depends on.
Low absorptive capacitySzulanski 1996Give the receiving line time and a person: one go-and-see visit and a trial run before any deadline.
Arduous relationshipSzulanski 1996Pair the source team leader with the receiving team leader for the first deployment.
Lessons not found at the right timeGAO-02-195Tag each lesson with the loss it fixes, so it appears when that loss appears on another line.
Lessons rarely usefulGAO-02-195Only deploy what was proven in data on the source line.
Intolerance of mistakesGAO-02-195Share failures and near misses as yokoten too. A lesson about what did not work saves the next line a trial.

Worked example: one changeover fix across six lines

Example data. The plant, the lines and the improvement are assumptions for a packaging plant. The labour values come from US government sources as marked, and the result is given as a range.

  • Six similar packing lines, each with a crew of 4 operators, running 250 days a year. Example data.
  • Each line changes over 3 times a day. Example data.
  • Line 1 runs a changeover kaizen and cuts the changeover from 40 to 25 minutes, proven over four weeks of data. Example data.

What the fix is worth on one line

Time freed per line: 15 minutes x 3 changeovers x 250 days = 11,250 minutes, or 187.5 line hours a year. With a crew of 4, that is 750 crew hours.

What those hours are worth depends on what you do with them:

  • Low end, freed time replaces straight-time hours (for example a shorter weekend shift): 750 x $30.43 = $22,822.50 a year. $30.43 is the October 2026 (preliminary) average hourly earnings of US manufacturing production and nonsupervisory employees in BLS Employment Situation Table B-8, wages only.
  • High end, freed time replaces overtime: overtime is paid at no less than time and a half under the FLSA, per US Department of Labor Fact Sheet #23, so 750 x $45.645 = $34,233.75 a year.
  • Zero if no hours are removed and the extra capacity is not sold. Be honest about this case: freed time only pays when something changes.

What yokoten multiplies it to

Deploying the fix to the five other lines multiplies the value by five: $114,112.50 to $171,168.75 a year, on top of line 1. That is the whole case for yokoten in one number: the kaizen was paid for once, and the return comes six times.

Now apply what the research says about stickiness. Suppose three receiving lines reach the full 15 minutes, and two reach only 7.5 minutes because their machines differ (example data). The five lines then deliver the equivalent of four full deployments: $91,290 to $136,935 a year. The gap of $22,822.50 to $34,233.75 is the price of copying without adapting, and it is why step 5 measures each line separately.

Copy, adapt or skip?Check each receiving line before you deploySame conditions?YesNoSameloss?YesNoCopyDeploy thestandard, thenverify in dataAdaptGo and see,trial, thenverify in dataShareFile the lesson,watch forthe lossSkipNo problemto solve here
A decision matrix for each receiving line: deploy only where the same loss exists, and adapt where conditions differ.

Book a demo and see OEE and losses per line and shift.

Yokoten compared with related practices

PracticeQuestion it answersScopeOutput
KaizenHow do we improve this process?One process or lineA better local standard
YokotenWhere else does this problem exist, and how do we spread the fix?Similar lines, shifts, plantsThe fix running everywhere it applies, verified
Standard workWhat is the current best way to do this job?One jobA documented method
Corrective actionHow do we stop this problem coming back?One nonconformity, and similar onesRoot cause removed
Lessons learned systemWhat did we learn, and where is it stored?Projects, organisationA searchable record
BenchmarkingWho does this better than us?Internal or externalA target and a source to learn from

Yokoten overlaps with each, but adds one thing the others often lack: the active search for every other place where the same problem exists, and the duty to check the result there.

Common mistakes

  • Sending the procedure, not the reason. This is causal ambiguity in practice. The receiving line cannot adapt what it does not understand.
  • Mandating by date. A top-down deadline produces paper compliance. Give the receiving line time to go and see and to run a trial.
  • Deploying unproven fixes. If the source line cannot show the before and after, there is nothing to spread yet.
  • Measuring the source line only. Each receiving line needs its own before and after. Averages hide the lines where it did not work.
  • Sharing only successes. The GAO found intolerance of mistakes was a barrier. A failed trial is a lesson worth spreading too.
  • Using different loss definitions on different lines. If one line calls a 4-minute stop a changeover and another calls it a breakdown, you cannot find the same problem across lines, let alone prove the fix.
  • Treating yokoten as a CI team project. The CI team can run the system, but line leaders own the transfer. Peer-to-peer is the point.

For plant, production and CI managers

This week

  1. List the last five improvements on your site that were proven in data. For each, write down which other lines have the same loss.
  2. Pick one with the largest number of receiving lines. Write the reason it works and the conditions it depends on on one page.
  3. Arrange one go-and-see visit for the leaders of two receiving lines.
  4. Record the baseline of the target loss on each receiving line: minutes, counts or scrap, for the last four weeks.

This quarter

  1. Agree common loss definitions across lines and sites, so the same problem has the same name everywhere.
  2. Add a yokoten line to every closed improvement: where else does this apply, who owns it there, and the date of the check.
  3. Track each deployment with before and after data per line, and report the result in money using a sourced labour or overtime rate.
  4. Share one failed trial a month as a lesson, with the same format as a success.
  5. Review which deployments stalled and why: skills, time, unclear cause or relationship. Fix the system, not the people.

Yokoten depends on seeing the same loss on many lines at once. OEE software records stops, changeovers, counts and scrap per line with the same definitions everywhere, and a light MES ties them to products and orders, so you can find where else a problem exists and prove the fix on each line.

How Fabrico fits

Fabrico is a cloud OEE and light MES platform. It shows OEE and losses by machine, line and shift, from the PLC or a Fabrico IoT device, with the reasons operators add on a tablet or phone. That gives you the same loss data on every line, which is what finding a problem elsewhere and proving a fix there needs.

On manual stations, AI cameras (an add-on) count output.

Book a demo and see downtime per line and shift.

What people ask next

What does yokoten mean?

Yokoten is short for yokotenkai, a Japanese term for horizontal deployment: applying the knowledge, lessons and techniques gained in one case to other cases. In lean manufacturing it means spreading a proven improvement to every line, shift or plant with the same problem.

How do I start yokoten across my lines?

Pick one improvement that was proven in data, write down why it works and what conditions it depends on, list the other lines with the same loss, send their leaders to see it running, and measure the same loss before and after on each line.

Who should own yokoten in my plant?

The CI team can run the system, but the transfer belongs to line leaders: the source team leader explains the fix and the receiving team leader adapts and proves it. Yokoten is peer-to-peer by design, not a head office instruction.

How do I know the fix worked on the other lines?

Measure the target loss on each receiving line for a few weeks before and after, with the same loss definitions as the source line. Report each line separately; an average hides the lines where it did not work.

Should every line be forced to adopt the same fix?

No. Copy it where the loss and the conditions match, adapt it with a trial where the conditions differ, share it as a lesson where the loss is absent, and skip it where neither applies.

What is the difference between yokoten and kaizen?

Kaizen improves one process. Yokoten takes that improvement and spreads it to every other process with the same problem, then checks the result there. Without yokoten, kaizen stays a local success.

Sources

All sources accessed on 14 November 2026. Dates are the publication dates shown on each page.

Track OEE and downtime in real time
See where lines lose time
Book a live demo.
See where lines lose time
See where lines lose time

Real Fabrico tiles, sample data.

By clicking the Accept button, you are giving your consent to the use of cookies when accessing this website and utilizing our services. To learn more about how cookies are used and managed, please refer to our Privacy Policy and Cookies Declaration