Process improvement methodologies: 9 methods compared

Process improvement methodologies are structured approaches for making work faster, cheaper, safer or more reliable. The main ones are PDCA, lean, kaizen, Six Sigma, Lean Six Sigma, TQM, theory of constraints, reengineering and Agile retrospectives. Each answers a different kind of problem.

This guide defines process improvement, then covers each method with its origin, where it fits and a short example. It ends with how to choose between them and how to combine them.

Key takeaways

Process improvement means studying how work gets done and changing it so the results get better.

Match the method to the problem: waste and delay, variation, a bottleneck, or a broken design.

Continuous improvement methodologies such as PDCA and kaizen make many small changes, while reengineering makes one large one.

Most organizations combine methods, using one as the cycle and others as tools inside it.

Every example in this guide is illustrative, not a documented company result.

What is process improvement?

Process improvement is the practice of analyzing how work flows and changing it so that quality, speed, cost or safety gets better. To define process improvement simply: understand the current process, find the problem, change it, and check that the change worked.

A process here means any repeatable chain of steps, from machining a part to approving an invoice. Continuous process improvement methodology adds one idea: the work never stops, so improvement repeats.

The continuous improvement page covers that ongoing habit in more detail.

Method matters less than the habit of managing improvement. A Census Bureau survey of about 35,000 US manufacturing plants, analyzed by Nicholas Bloom and colleagues in the American Economic Review in 2019, found that structured management practices, such as monitoring performance and setting targets, account for more than 20% of the variation in productivity between plants.

The nine methods at a glance

Each line gives the problem the method fits, the data it needs, a typical length, the scale of change and who usually runs it. Lengths are typical ranges, not rules.

PDCA: small, testable changes to any process. A simple before and after measure. Days to a few weeks. Small changes. The team doing the work.

Kaizen: everyday problems at the workplace. Observation and simple counts. Daily, or a 3 to 5 day event. Small to medium changes. Frontline teams, with a facilitator for events.

Lean: waiting, rework and excess stock. Process maps with lead and touch times. Weeks per project, years as a program. Medium changes across a value stream. Leaders and cross-functional teams.

Six Sigma: variation and defects. Reliable measurement data and statistics. Several months per DMAIC project. Medium changes. Trained Green Belts and Black Belts.

Lean Six Sigma: waste and variation together. Both kinds of data. Weeks to months per project. Medium changes. Belt-trained project leaders.

TQM: quality culture across the organization. Customer and quality measures. Years. Broad change in management practice. Senior leaders and every function.

Theory of constraints: one step that limits output. Throughput at each step. Weeks. Focused changes at the constraint. Operations managers.

Reengineering: a process broken by design. Process cost and cycle time. Months. Large, radical change. A senior sponsor and a dedicated team.

Retrospectives: teams working in short iterations. Team observations and flow metrics. Every 1 to 4 weeks. Small changes. The team itself.

Where each method tends to go wrong

PDCA and kaizen: the continuous improvement core

PDCA is a four-step cycle of plan, do, check and act, and kaizen is the habit of running such cycles every day. Together they form the base of most continuous improvement methodologies.

Walter Shewhart described a three-step cycle of specification, production and inspection in his 1939 book Statistical Method from the Viewpoint of Quality Control. W. Edwards Deming adapted it into a four-step cycle in lectures in Japan in 1950, and Japanese managers recast that as plan, do, check, act. That is why it is called both the Shewhart cycle and the Deming cycle. Deming later preferred PDSA, with study in place of check. Masaaki Imai's 1986 book Kaizen: The Key to Japan's Competitive Success brought kaizen to a wide audience.

Good fit: small, frequent changes by the people doing the work. Say a team trials a new tool layout for one week, checks the walking distance and then adopts or drops it. See the PDCA method and the kaizen event guide.

Lean: removing waste and improving flow

Lean is a way of improving processes by removing waste so that value flows to the customer without delay. It grew out of the Toyota Production System.

John Krafcik used the term lean in a 1988 article. Womack, Jones and Roos popularized it in their 1990 book The Machine That Changed the World. Womack and Jones followed with Lean Thinking in 1996.

Good fit: processes with long waits, rework, excess stock or unclear flow. Say an order sits in three queues before shipping. A value stream map shows the touch time is a small part of the lead time. The lean tools library lists the methods used.

Small manufacturers can get outside help with lean projects. Clients of the Manufacturing Extension Partnership, a NIST program that supports small and medium-sized manufacturers, reported $15 billion in new and retained sales and $2.6 billion in cost savings in fiscal year 2024.

Six Sigma, DMAIC and Lean Six Sigma

Six Sigma is a data-driven method for reducing variation and defects, usually run as DMAIC projects: define, measure, analyze, improve and control. Lean Six Sigma combines it with lean waste removal.

Six Sigma was developed at Motorola in the 1980s, and engineer Bill Smith is often credited with its origin. Michael George's 2002 book Lean Six Sigma: Combining Six Sigma Quality with Lean Speed helped establish the combined approach.

Good fit: a stable process that produces inconsistent results and where data can be collected. Say a filling line gives uneven volumes, and a DMAIC project finds which factor drives the spread. See DMAIC and lean vs Six Sigma.

Total quality management (TQM)

Total quality management is an organization-wide approach in which every function and every employee takes responsibility for quality and customer satisfaction. It focuses on culture and management practice more than on a single tool.

The ideas grew from the work of Deming, Joseph Juran and Armand Feigenbaum, and the term became common in the 1980s. Good fit: organizations that want a shared quality culture, with leadership commitment and customer focus as the base.

In the United States, the Baldrige Performance Excellence Program recognizes organizations for this kind of quality management. A 2011 economic evaluation for NIST by Albert Link and John Scott estimated its benefit-to-cost ratio at 820 to 1, comparing the benefits reported by 273 award applicants from 2007 to 2010 with the program's cost.

An example: a hospital laboratory sets quality goals for every team, reviews customer complaints monthly and trains staff in basic problem solving.

Theory of constraints

The theory of constraints says that every system has one limiting step, and improving anything else will not raise output. You find the constraint, get the most from it, then move to the next one.

Eliyahu Goldratt introduced it in his 1984 novel The Goal. Good fit: a process with a clear bottleneck, where work piles up in front of one step.

Say a print shop's finishing machine handles 50 jobs a day while upstream steps could do 80. Adding capacity upstream changes nothing, but a faster finishing step raises output.

Business process reengineering

Business process reengineering is a radical redesign of a process from a blank sheet, aimed at a large jump rather than gradual gains. It questions why the work is done at all.

Michael Hammer set out the idea in his 1990 Harvard Business Review article Reengineering Work: Don't Automate, Obliterate. He and James Champy expanded it in the 1993 book Reengineering the Corporation.

Good fit: a process that is broken by design, such as a ten-step approval chain built around old paperwork. The risk is disruption, so it suits rare, large changes. Compare it with the small steps in kaizen.

Agile retrospectives

An Agile retrospective is a regular team meeting that reviews how the work went and picks one or two changes to try next. It is process improvement on a short cycle for software and knowledge work.

The Agile Manifesto principles of 2001 say the team reflects at regular intervals on how to become more effective. Esther Derby and Diana Larsen's 2006 book Agile Retrospectives gave teams a practical format.

Good fit: teams working in short iterations. Say a team notices reviews wait two days, and agrees to pair on reviews for the next sprint. See the retrospective tools category.

How to choose a process improvement methodology

How to combine methodologies

Combine methodologies by using one as the improvement cycle and the others as tools inside it. PDCA often works as the cycle, with lean mapping to find waste and DMAIC statistics to confirm causes.

A practical sequence: map the process, find the constraint, run a focused kaizen event on it, then use DMAIC if variation remains. Reserve reengineering for processes that small changes cannot fix.

Keep the combination simple. One shared language and one improvement cycle are better than three methods run by three separate teams.

A combined example

An illustrative example. A distributor's order-to-ship lead time averages 6 days, and customers want 3.

Lean first: a value stream map shows about 40 minutes of touch time inside the 6 days, with most of the wait at credit checks. Theory of constraints next: the credit check is the constraint, because one analyst clears about 60 orders a day while 80 arrive.

The team runs a kaizen event on the credit step and sets an auto-approval rule for repeat customers within their limit, which removes about half the checks. PDCA keeps it honest: the rule is trialed for two weeks and checked for bad-debt cases before it becomes standard.

Lead time falls to about 3.5 days. Order accuracy still varies by warehouse shift, so a DMAIC project is chartered for that next. Each method answered a different question, and one PDCA cycle held them together.

Process improvement examples by setting

Steps that apply to any method

Common process improvement mistakes

Frequently asked questions