The continuous improvement process steps come down to six moves: map and measure the current process, find the root cause, design a fix, pilot it, check the pilot against your baseline, and standardize what worked. Then you repeat the loop on the next problem. This is the Plan-Do-Check-Act (PDCA) cycle with the planning and acting phases spelled out. The four-, five-, and seven-step versions you will see elsewhere are the same loop counted differently.
Most improvement efforts do not fail for lack of ideas. They fail because teams skip steps. They jump to a solution before they know the cause, roll a change out to the whole plant without a pilot, or never write the new method down, so old habits return within weeks. Each skipped step feels like time saved. Each one is really where the effort gets wasted.
This guide walks you through every step with specific actions, copy-ready templates, and a worked example. Lean Six Sigma Experts has worked since 2011 as an engineering-based consulting, training, and recruiting firm, so the method leans on measurement over opinion. By the end, you can run a full cycle on one process in your own workplace.
Why a defined process beats ad hoc fixes
A defined process beats ad hoc fixes because it forces you to prove the problem, the cause, and the result with data instead of opinion. The continuous improvement process steps give your team a shared sequence, so nobody argues about solutions before everyone agrees on the problem.
What continuous improvement means in practice
Continuous improvement is the ongoing practice of making small, tested changes to how work gets done, then locking in the ones that work. The Japanese word Kaizen means "change for the better" and names the philosophy behind it. PDCA, developed from the work of Walter Shewhart and popularized by W. Edwards Deming, is the working cycle that turns the philosophy into action.
Small does not mean trivial. Cutting a changeover by four minutes sounds minor until it repeats across every shift and line, every day. Gains compound. ISO 9001 also asks organizations to continually improve their quality management system (clause 10.3), so auditors expect to see a repeatable method rather than good intentions.
Four, five, six, or seven steps
Every version describes the same loop. The difference is how finely the early diagnosis work is split up. This table shows how the common counts line up.
| Version | Steps | Notes |
|---|---|---|
| PDCA | Plan, Do, Check, Act | Shortest model. "Plan" hides the diagnosis work. |
| 5-step | Identify opportunity, plan, implement, evaluate, standardize | Common in business and service settings. |
| 6-step (used here) | Map and measure, find root cause, design fix, pilot, check, standardize | Steps 1 to 3 are Plan, step 4 is Do, step 5 is Check, step 6 is Act. |
| 7-step | Splits diagnosis further, such as select the problem, measure it, analyze causes | Common in Kaizen and Lean training. Wording varies by author. |
Pick one count and use it consistently across your organization. We use six because it separates the three places where teams cut corners: measuring the baseline, verifying the cause, and standardizing the result.
Every version is the same loop, so what matters is that you never skip the diagnosis, the pilot, or the standard.
What to have in place before step 1
Preparation takes an hour, and it prevents most of the stalls that appear later. Before you start the first of the continuous improvement process steps, confirm these basics:
- One process with a named process owner, not a whole department.
- A metric you can already measure, or a plan to start measuring it.
- Time with the people who do the work, since they know where the process breaks.
- A sponsor with authority to approve a pilot and release resources.
- A calendar slot for a short weekly review, so the cycle keeps moving.
Steps 1 and 2. Find the problem and its cause
The first two continuous improvement process steps are diagnosis. You measure the current process, then find out why it underperforms. Finish both before anyone proposes a solution.
Step 1. Map and measure the current process
Walk the process from first input to final output with the people who run it. Draw what actually happens, not what the written procedure says. A simple flowchart works, and a SIPOC diagram (suppliers, inputs, process, outputs, customers) works better for processes that cross departments. Gaps between the documented and the real process are often the first findings.
Next, choose one metric that reflects the pain, such as defect rate, cycle time, or first-pass yield. Collect a baseline across enough normal operation to see its usual variation, which is typically a few weeks. If several problems compete for attention, rank causes with a Pareto chart, because a few categories usually account for most of the loss. Then write a problem statement with this template:
In [process], [metric] is [current value] against a target of [target value],
measured from [start date] to [end date] at [location].
This costs [impact in dollars, hours, or customer complaints].
Step 2. Find the root cause
Use the 5 Whys to move from symptom to cause. Keep asking why until you reach something you can fix permanently. Here is a short example from a packaging line where 7% of cartons need relabeling:
- Why are cartons relabeled? The labels land crooked.
- Why do they land crooked? The applicator head shifts during a run.
- Why does it shift? Its mounting bolt works loose.
- Why does the bolt loosen? It has no locking hardware and is not on the maintenance checklist.
- Why is it missing from the checklist? The line was installed without a preventive maintenance plan.
The root cause is the missing maintenance plan, not operator carelessness. Five is a rule of thumb, not a quota. When several causes may contribute, build a fishbone diagram and group causes by method, machine, material, people, measurement, and environment.
Then verify the cause with data. Compare defect rates by shift, machine, or supplier and see whether the pattern matches your theory. This is where an engineering mindset pays off, because a cause you have not tested is only a hypothesis.
An unverified root cause is a guess, and a fix built on a guess is the most common way improvement efforts waste money.
Steps 3 and 4. Design the fix and test it small
Steps 3 and 4 turn the diagnosis into action without betting the whole operation on an untested idea. In the continuous improvement process steps, this is the Plan-to-Do handoff, and a small pilot is what keeps it safe.
Step 3. Design the solution and set a target
Gather the people who do the work and list every solution that addresses the verified cause. Score each option on impact and effort, then pick the one with the best ratio. Favor fixes that remove the cause over fixes that ask people to remember something. Locking hardware and a maintenance checklist beat a reminder email, because error-proofing does not depend on memory.
Set a target tied to your baseline, such as "reduce relabeling from 7% to under 2% during the pilot." Then build an action plan. Every action needs an owner, a due date, and a way to confirm it is done:
| Action | Owner | Due | Evidence of completion |
|---|---|---|---|
| Install locking hardware on applicator | Maintenance lead | Day 3 | Photo and work order closed |
| Add bolt torque check to daily checklist | Line supervisor | Day 3 | Checklist revision signed |
| Brief operators on the new check | Shift lead | Day 5 | Sign-in sheet |
Step 4. Pilot the change on a small scale
Run the change on one line, one shift, or one site before you touch anything else. A pilot limits the damage if the idea fails and gives you a clean comparison against the baseline. Change only the variable you are testing, and measure the same metric the same way you did in step 1. Write the pilot down first:
Pilot: [change being tested]
Scope: [line / shift / site]
Duration: [start date] to [end date]
Metric and method: [metric], measured by [who] using [tool] every [interval]
Success threshold: [target value]
Stop rule: end early if [safety, quality, or customer risk condition]
Review points: [dates for check-ins]
Hold a short daily huddle during the pilot so problems surface fast. If you can, keep a similar line running the old method as a control. When both lines see the same shift in results, something other than your change is at work.
Steps 5 and 6. Check the results and lock them in
Steps 5 and 6 decide whether the change is real and then make it permanent. This is where the continuous improvement process steps separate lasting gains from short-lived ones.
Step 5. Check the pilot against your baseline
Compare pilot results to the baseline using the same metric, the same method, and a similar time window. Plot the data on a run chart or control chart so you can see variation rather than a single before-and-after number. A difference that sits inside normal variation is not evidence of improvement. When the gap is close, practitioners use a statistical test to confirm it, and that is where Six Sigma training pays off.
Also look for side effects. Did the fix move a problem to another step, slow the line, or create a safety concern? Then decide using this table:
| Result | Decision | Next move |
|---|---|---|
| Target met, no side effects | Adopt | Go to step 6 |
| Partial improvement | Adapt | Return to step 3 and refine the fix |
| No improvement or harm | Abandon | Return to step 2 and recheck the cause |
Step 6. Standardize and start the next cycle
Standardizing means the new method becomes the only documented way to do the work. Update the standard work or SOP, train everyone who touches the process, and add visual cues such as checklists, labels, or shadow boards. Then create a control plan so you can see drift early. It needs five things:
- The metric to watch and its acceptable range.
- The person who reviews it.
- How often they review it.
- What to do when the metric leaves its range.
- The date of the first audit.
Share the result with the team that ran the pilot and name what they contributed. Recognition keeps the next round of ideas coming. Then return to step 1 with the next-largest problem from your Pareto chart.
A fix that is not written into standard work is a loan, and the old habits will collect on it.
A worked example and the tools behind each step
A complete cycle looks smaller than people expect. The example below uses round, illustrative numbers to show how the continuous improvement process steps connect, followed by the tools that support each one.
One full cycle on a single process
Take a packaging line with a 7% relabel rate. This illustrative example follows the six steps in order.
| Step | What the team did | Output |
|---|---|---|
| 1. Map and measure | Mapped the line, tracked relabels for three weeks | Baseline of 7%, problem statement |
| 2. Find root cause | Ran 5 Whys, compared defects by machine | Verified cause: no maintenance plan for applicator |
| 3. Design fix | Chose locking hardware plus a daily torque check | Action plan, target under 2% |
| 4. Pilot | Applied the fix to one line for two weeks | Pilot data, same metric |
| 5. Check | Compared pilot to baseline on a control chart | Relabels fell outside normal variation, adopt |
| 6. Standardize | Updated checklist, trained all shifts, set weekly audit | New standard work, control plan |
Tools to use at each step
You do not need every tool. Match the tool to the question the step asks.
| Step | Helpful tools |
|---|---|
| 1. Map and measure | Flowchart, SIPOC, data collection sheet, Pareto chart |
| 2. Find root cause | 5 Whys, fishbone diagram, stratified data comparison |
| 3. Design fix | Impact and effort matrix, error-proofing, action plan |
| 4. Pilot | Pilot plan, daily huddle, control line |
| 5. Check | Run chart, control chart, hypothesis test |
| 6. Standardize | SOP, standard work, visual management, control plan |
Mistakes that stall the cycle
Most stalled efforts trace back to a short list of habits. Watch for them early, because each one is cheaper to fix in week two than in month six.
- Starting with a solution someone already likes, then hunting for a problem to fit it.
- Measuring the baseline informally, so the result cannot be proven later.
- Piloting everywhere at once, which removes your comparison group.
- Declaring victory without a control plan.
- Running one big project and calling it continuous improvement, when the point is the repeated loop.
Continuous improvement works when the loop repeats, so favor many small, proven cycles over one heroic project.
Start with one process and one loop
The continuous improvement process steps are simple to list and hard to follow without shortcuts. Map and measure, find the verified root cause, design the fix, pilot it, check against your baseline, and standardize. Then pick the next problem and run the loop again. Whether you count four, five, six, or seven steps, the discipline is the same.
Choose one process this week, name its owner, and write the problem statement from step 1. If you want an engineering-based partner to help run your first cycle, train your team, or build an improvement program across sites, talk with the Lean Six Sigma Experts team.