The Six-Step Kaizen Procedure
Kaizen is not improvisation. It is a disciplined procedure grounded in the scientific method — observe, hypothesize, test, confirm — expressed as six repeating steps that together form the kaizen (改善) cycle.
Telling people to "do kaizen" without giving them a procedure produces activity without results. The procedure is what turns awareness of waste into actual improvement: it channels observation into analysis, analysis into ideas, ideas into a plan, the plan into action, and action into a confirmed new standard. Each step builds on the one before. Skip one and the chain breaks.
By the end of this section, you should understand:
- why kaizen follows the same underlying logic as problem solving and the scientific method,
- why step counts vary across organizations — and what remains constant,
- how 6-, 8-, and 12-step patterns all resolve to the same four PDCA phases,
- what each of the six steps accomplishes before handing off to the next.
The kaizen procedure is a six-step cycle anchored in PDCA. The steps repeat without end: each confirmed improvement raises the standard, which then becomes the baseline from which the next cycle begins. The cycle does not close; it spirals upward.
1The six-step cycle
The six steps are: Discover the improvement potential, Analyze the current situation, Generate original ideas, Develop the implementation plan, Implement the plan, and Evaluate the new method. They form a loop — Step 6 leads back to Step 1, because confirming one improvement always surfaces the next.
Two things distinguish kaizen from problem solving, even though the underlying logic is the same. Problem solving starts from a deviation: a known standard exists, something has gone wrong, and the goal is to close the gap back to standard. Kaizen often starts before any standard is violated — from a leader's eye for waste, from a challenge to improve cost, safety, or lead time even when today's performance is acceptable. The trigger differs; the discipline is the same.
Both are applications of the scientific method: establish the current condition, form a hypothesis about how to improve it, test the hypothesis under real conditions, and check whether the results hold. PDCA — Plan, Do, Check, Act — is the compact name for that same pattern.
The six steps run in a fixed order and loop back to Step 1 after Step 6. Confirming that a new method works raises the standard — and that raised standard becomes the starting point for the next cycle.
What to notice: the dashed arrow from Step 6 back to Step 1 is not a failure path — it is the intended path. Kaizen does not conclude; it recycles. The 改善 (kaizen) character at the centre is not decoration: it is the name of the discipline being practiced.
2Why the step count varies
If you have encountered kaizen training elsewhere, you may have seen eight steps instead of six, or twelve instead of eight. All of them are legitimate — and they all describe the same underlying logic.
The reason for the variation is straightforward: kaizen on a work method (standardized work) emphasizes different activities than kaizen on a flow (material and information movement between processes), which in turn emphasizes different activities than kaizen on equipment or machines (availability, performance, quality losses). Each domain needs slightly different sub-steps. Organizations that teach kaizen across all three domains tend to expand the count to accommodate the additional activities; organizations focused on one domain can compress. No single agreed count ever emerged because there was no single agreed scope.
What does not vary is the four-phase structure of PDCA. Every kaizen step list, regardless of count, allocates its steps to the same four phases: Plan (define the problem and the target, analyze the current state, generate and select ideas), Do (implement the chosen improvement), Check (verify the result against the original target), and Act (standardize what worked; surface the next problem). The six-step version used in this guide covers those four phases; so does every eight-step or twelve-step version you will encounter. The differences are in granularity, not in logic.
| Version | Plan | Do | Check | Act |
|---|---|---|---|---|
| 6 steps | 1 Discover the potential 2 Analyze the current situation 3 Generate original ideas 4 Develop the implementation plan | 5 Implement the plan | 6 Evaluate the new method | (6→1) Update the standard; begin the next cycle |
| 8 steps (typical flow focus) | 1 Select the theme 2 Grasp the current state & set the target 3 Analyze the causes 4 Plan countermeasures | 5 Implement countermeasures | 6 Confirm the effect 7 Standardize & prevent recurrence | 8 Review the process & plan next steps |
| 12 steps (equipment / TPM focus) | 1 Select equipment theme 2 Form the project team 3 Grasp current losses (OEE baseline) 4 Set the target 5 Identify causes (PM analysis) 6 Plan restoration & improvement 7 Implement planned maintenance changes | 8 Confirm and trial-run 9 Evaluate results vs target | 10 Establish visual standards 11 Replicate horizontally | 12 Review results; plan the next project |
Three common kaizen step-count conventions, organized by PDCA phase. The 6-step version is the one taught in this guide and is sufficient for work-method kaizen at the team-leader level.
What to notice: the Plan phase absorbs the most steps in every version — because most of the analytical work happens there. Do is always just one or two steps. If a version you encounter elsewhere looks unfamiliar, map its steps to these four columns and you will find the logic unchanged.
The version of kaizen a factory teaches tends to reflect what its biggest improvement problems are. A factory running TPM will teach twelve steps because equipment restoration and horizontal replication matter enough to name explicitly. A factory focused on standardized work will teach six. The step count is a clue to what the organization considers the hard part — not evidence that one version is more rigorous than another.
3What each step does
The following is a brief account of each step before the deep dives in Sections 4 through 8. Think of it as a map of the territory — enough orientation to understand how the pieces connect.
Establish what you will try to improve and why. The need is sometimes obvious — a safety issue, a quality defect, a cost target — and sometimes requires looking carefully to surface waste that is taken for granted. Without a clear target and a standard to compare against, there is no way to measure whether the kaizen actually worked.
Generate an accurate picture of how work is actually done right now, based on facts observed at the actual workplace with actual objects at the actual time — not based on memory, assumption, or what the standard document says. Conjecture at this step contaminates every step that follows.
Produce as many candidate ideas as possible before evaluating any of them. Judging ideas while generating them is the most common way to kill the best answer early. Breadth at this step is the resource that Step 4 will sort and test.
Review the ideas from Step 3, assess feasibility, estimate cost and difficulty, consider likely results, and draw up a specific plan with named responsibilities and dates. Give priority to work-method kaizen over equipment solutions — the improvement that requires no capital is the one worth finding first.
Execute the plan. This requires involving all affected team members and departments — not just notifying them, but explaining why the change is being made, asking for their input, and arranging any retraining. The best improvement plan fails if the people doing the work do not understand it or were not part of it.
Compare results to the target set in Step 1. Check daily during the first period; watch for unexpected problems and devise countermeasures. Standardize whatever works. Then begin the next cycle — because a confirmed improvement that raises the standard immediately reveals the next opportunity waiting above it.
One caution on Step 6: evaluation is not the end of kaizen — it is the return to Step 1. The standard that existed before the kaizen was the baseline; the confirmed new method becomes the new baseline. That new baseline is what the next cycle of observation will scrutinize. Kaizen without standardization produces temporary results. Standardization without the next kaizen cycle produces stagnation. The procedure is designed so that neither mistake has a natural stopping point.
Skipping Step 2 because the problem seems obvious. The pressure to move directly from "we have a waste" to "here is my idea" is strong, especially when the team leader already has a solution in mind. But the analysis in Step 2 is what separates a real cause from a plausible-sounding one. An idea selected without a thorough analysis of the current situation often solves the visible symptom and leaves the root condition untouched.
Section summary
The kaizen procedure is a six-step cycle grounded in the scientific method and expressed as PDCA. Steps 1 through 4 are the Plan phase — discover the potential, analyze what is actually happening, generate ideas, develop the plan. Step 5 is Do. Step 6 is Check, with the Act phase built into the transition: standardize what worked, then let the confirmed new standard begin the next cycle at Step 1.
Step counts vary across organizations — six, eight, even twelve — because kaizen on work methods, kaizen on flow, and kaizen on equipment each require different analytical sub-steps. The count is not the issue; the underlying PDCA logic is. Any version you encounter, mapped onto Plan-Do-Check-Act, will look the same as any other.
Sections 4 through 8 work through each step in depth. The key thing to carry into those sections: every step depends on the one before it. An accurate current-state analysis (Step 2) is only possible after a clear target is set (Step 1). A sound plan (Step 4) is only possible if the ideas it selects from are genuinely broad (Step 3). Skipping any step does not speed up the kaizen — it undermines it.