kaizen in practice · by shop type
D-KAI-07 | First published 2026-07-18 | Revised 2026-07-18 | Primary sources: 5
Disclosure: Gemba Works Lab is supported by advertising. This page contains no affiliate links; if any are added in the future they will be clearly marked (per the FTC Endorsement Guides).
Kaizen's reference examples come from repetitive manufacturing — the same part, the same line, thousands of cycles — and the method's most famous records were set there: Toyota's suggestion system was processing 220,000 employee ideas a year by 1973, with more than 70% adopted[4]. Most American small manufacturers do not look like that. U.S. machine shops alone employ 265,030 people[1], and the typical shop in that industry runs high-mix, low-volume work: this week's job is not last week's job. The question this guide answers for the job-shop owner reading repetitive-plant case studies: which parts of the kaizen toolkit are portable across that gap, which need translation, and which will waste your quarter?
KEY POINTS
The two shop types waste differently. A repetitive plant wastes in-cycle: seconds of extra motion repeated ten thousand times, a line imbalance that idles one station forever. A job shop wastes between cycles: the machine sits still while someone hunts a fixture, re-reads a traveler, re-programs, re-quotes, or waits for an answer. Same seven wastes, inverted proportions. That inversion is why a job-shop owner who copies a repetitive plant's kaizen target list — cycle-time shaving, line rebalancing — sees small returns and concludes the method does not apply. The method applies; the target list was borrowed from the wrong shop.
The wage stakes are the same either way. A machinist in NAICS 332710 earns a median $23.62 an hour[1]; an hour of that machinist watching a spindle idle during a two-hour setup costs the same whether you call your shop "job" or "repetitive." The difference is only where the recoverable hour hides.
| Tool | In a repetitive plant | In a job shop | Transfer verdict |
|---|---|---|---|
| Teian (suggestions) | Volume engine — Toyota's system grew from 789 ideas in 1951 to 220,000/year by 1973[4] | Identical mechanics: form, board, 72-hour response. Machinists see more process variety, so suggestions skew toward fixtures, tooling and information flow | Unchanged |
| 5S | Anchors visual control on the line | Arguably pays faster: search time is a dominant between-cycle waste when every job needs different tooling | Unchanged |
| Setup reduction (SMED-style) | One changeover pattern, deeply optimized | The premier target — more changeovers per week than any repetitive plant; reported event results concentrate here (50–70% cuts)[2] | Unchanged — and higher leverage |
| Kaizen events | Aimed at a cell or line segment | Same three-phase mechanics (prepare, execute, follow up)[3], aimed at a recurring process: the changeover family, the quoting loop, first-article inspection | Unchanged mechanics, different targets |
| Standard work | One best sequence per station, timed | Standardize the wrapper, not the part: setup checklist per machine family, traveler format, tool-return rule, program-naming convention | Needs translation |
| Takt / line balancing | Core discipline | Largely inapplicable: demand is lumpy and routings differ per job. Flow gains come from scheduling and queue rules instead | Mostly skip |
| Single-family kanban | Pull between fixed stations | Works only for the boring stable stuff: perishable tooling, hardware, shop supplies. Use it there; do not force it onto job flow | Partial |
The repetitive plant standardizes the product cycle. The job shop standardizes everything that recurs even though the product changes: how a job is quoted, how a traveler is written, how a setup begins and ends, how first articles are checked, where tools go when a job closes. Each of those is a process your shop runs dozens of times a week — more cycles than many repetitive stations see — and each can carry a one-page standard, a baseline measurement, and a kaizen target. The EPA's event framework does not care that the "process" is quoting rather than welding; the prepare-execute-follow-up phases and the monthly review cadence apply identically[3]. Our 5-day event checklist (D-KAI-03) works for a changeover-family event without modification; run the scoping step against a recurring process, not a part number.
There is also a job-shop advantage nobody advertises: because every machinist touches varied work, your crew's suggestion pool is wider than a repetitive line's. The same participation mechanics the public guidance documents — workers surface the problems they know best[5] — produce fixture ideas, tooling standards, and traveler fixes that a single-part operator would never see. The cost side is unchanged from our 20-person budget guide (D-KAI-06): forms, a board, and supervisor hours.
The transfer runs the other way too. Small repetitive plants — the 40-person shop running the same four products — often over-invest in event theater and under-invest in the daily routine, precisely because their processes look "already standard." The Toyota record is the counterargument: the system's yield came from suggestion volume sustained over decades, from 789 ideas in year one to hundreds of thousands a year two decades later, with adoption rates climbing from 29% to 76% as the response machinery matured[4]. The cadence choice between concentrated events and that daily drumbeat — and what each produces on the record — is the subject of Kaizen Events vs Daily Teian Systems (D-KAI-08).
Two boundary cases, whichever shop you run. If your improvement problem is dominated by process variation you cannot see from the floor, the statistical toolset wins — the boundary is mapped in Kaizen vs Six Sigma (D-KAI-04). And if the real constraint is that your best machinist retires next year with the setup knowledge in their head — a risk the machinist pipeline numbers make concrete[1] — capture comes before improvement: start with video SOPs (D-SKT-04) and the Skills Transfer section.
next step
Whichever shop type you run, the sequence is the same: one cell, one metric, three habits — costs, timeline and failure modes are laid out in the pillar guide.
Open the Complete Kaizen Implementation Guide (D-KAI-P01) Related: 10 kaizen patterns (D-KAI-02) · what kaizen means (D-KAI-01) · restarting after a failed rollout (D-KAI-05) · full section indexLast updated: 2026-07-18 | Primary sources referenced: 5 | Framing labeled as editorial synthesis where no measured statistic exists. Spotted an error? Tell us via the contact page.