Kaizen Continuous Improvement: How to Prove It Works

Kaizen continuous improvement means making small, ongoing changes with the people who do the work. The Japanese word 改善 means “change for the better.” The method became familiar in Western management through Masaaki Imai’s 1986 book and its account of the Toyota Production System.

Kaizen helps teams improve, but improvement alone does not show whether a change worked. Without a baseline, you may agree that a process feels better without knowing how much it changed. That makes it harder to defend the gain or spot when the process drifts back.

The method rests on two ideas: start from a defined standard, and involve the people at the gemba, where the work happens. This guide covers daily kaizen and kaizen events, how kaizen bursts connect a value stream map to an improvement plan, why programmes lose momentum, and how an event log can make results easier to measure.

What does kaizen mean?

Kaizen is a way to improve work through small, continuous changes. Two ideas make the method practical.

Start with a standard. You need a defined way of doing the work before you can tell whether it has improved. Kaizen changes the standard, documents the new way, and uses it as the basis for the next improvement. Without a standard, a change can remain an individual workaround and disappear when that person moves on.

Involve the people at the gemba. Gemba means the place where the work happens. The people doing the work see friction that may not appear in a report, and they have a direct role in maintaining a better way of working.

Tools such as 5S, PDCA, poka-yoke, kanban and andon can support these ideas. They are not kaizen itself. The core practice is to make a change against a baseline, check what happened, and keep the change if it helped.

Toyota’s distinction between innovation and kaizen is useful: innovation makes a large leap, while kaizen makes smaller steps with broad participation. Many improvements involve a change to a policy, a system setting or a form, rather than a major project.

How does daily kaizen differ from a kaizen event?

Daily kaizen and kaizen events serve different purposes. Daily kaizen keeps improvement part of the team’s work. An event brings people together to focus on a specific problem.

A common mistake is to run several events and treat them as the whole programme. Events can produce visible changes, but they do not create an ongoing habit by themselves. When the event ends, your team still needs to own the standard and continue looking for small improvements.

What is kaizen continuous improvement in value stream mapping?

A kaizen burst is a starburst symbol on a current-state value stream map. It marks the specific step where you want to make an improvement. In kaizen value stream mapping, each burst connects an observed problem to a proposed change and an expected result.

Use this rule to make a burst actionable:

• One step: Attach it to a specific step, not the process as a whole.
• One change: State what you plan to change, not just what you want to improve.
• One expected number: Estimate the effect in measurable terms.

“Reduce waiting” is too broad. “Move the credit check so it runs in parallel with the address check, aiming to remove two days of queue at this step” identifies a step, a change and an expected effect.

The bursts on a current-state map form an improvement plan for the future-state map. Each proposed change should have a reason and an expected result. Without a number, the list is harder to assess and may reflect who spoke last rather than where the process needs attention.

Value stream mapping explains the map and how current and future states fit together. The key question for each kaizen burst is how you will measure its effect. For more on the notation, see the value stream mapping documentation.

Why do kaizen programmes lose momentum?

Four common problems make it difficult to sustain improvements.

There is no baseline. Your team changes a step but does not measure it first. Later, when someone asks whether the change helped, the answer depends on memory and anecdotes.

The new standard is not documented. A change may work at first, but the process can drift if the new way is not reflected in the standard or system configuration.

Measurement takes too much effort. A manual time study can require sampling, stopwatches and agreement on how to measure the work. If the effort seems greater than the expected benefit, your team may skip the baseline. That leaves you without evidence to support the change.

The target is chosen by complaint, not by impact. Without data, your team may focus on the most visible or frustrating step. It may not be the step causing the most delay across the process.

These problems do not mean your team lacks commitment. They often mean the process is difficult to see and measure, so it is hard to choose a target or confirm the result.

How can an event log support kaizen?

If work happens in a system, that system may record events such as a case identifier, an activity and a timestamp. Together, those records can form an event log. Process mining uses event log data to show how the process ran, which can support the measurement work around kaizen.

Build a baseline from process data. You can examine cycle times, rework loops, process variants and waiting between activities. The median and the spread can show how typical and slower cases differ. The 90th percentile can help you understand the slowest cases, where long waits may be concentrated.

Choose a target based on observed delays. Comparing waiting times across steps can help you decide where to focus, rather than relying only on the loudest complaint.

Compare before and after using the same measure. If you use the same definition and scope, you can compare the process before the change with later data. The comparison still depends on choosing a fair period and accounting for changes in the process or case mix.

Check whether the process follows the new standard. If you define rules for the new way of working, conformance checking can help identify cases that do not follow them. That can make drift visible sooner, so you can investigate it while the change is still recent.

Process mining does not replace kaizen. It can replace some of the manual measurement work that makes it difficult to prove whether a kaizen change helped. Process mining can help you examine event log data and compare how work flows through your process.

Learn what you need to create a process mining event log.

How does kaizen compare with Lean and Six Sigma?

Kaizen, Lean and Six Sigma can support the same improvement effort, but they focus on different questions.

Many organisations combine Lean and Six Sigma with a kaizen habit. The approaches differ, but all need a way to measure the process and check whether a change had the intended effect. Read what Six Sigma is to compare the methods in more detail.

How can you run a kaizen cycle in a quarter?

Use one process, one step and one measure to keep the cycle focused.

1. Choose a process with an event log. Look for work that repeats regularly and runs in a system that records activities and timestamps. If the work is not recorded in a system, start by working out how you will measure it.
2. Baseline the flow before choosing a step. Measure waiting and working time across the process. The data may point to a different target than the one people initially expect.
3. Write the kaizen burst. Name one step, one change and one expected effect with a number. Assign an owner and a date.
4. Make the change and update the standard. Document the new way of working or update the relevant form or system configuration. A change that exists only in one person’s routine is harder to sustain.
5. Measure the result using the same definition. Compare the same measure across a suitable period. If the result did not improve, record that and decide what to test next. A measured change that did not work still gives you useful information.
6. Check conformance and repeat. Define how you will check whether the new standard is being followed. Then use the next cycle to identify another improvement.

Keep the scope to one process for each cycle. Ask the process owner to take part in measurement and review, so your team can continue checking results without relying on a specialist for every change.

What should you do next?

Start by measuring one step before you change it. If the work is recorded in an event log, you can use that data to build a baseline and decide which kaizen burst to test.

See the kaizen burst notation on the value stream page.