DMAIC: Define, Measure, Analyze, Improve, Control

DMAIC drives costly variation from manufacturing and business processes.  The five steps in DMAIC are   Define, Measure, Analyze, Improve, and Control.   As the backbone of the Six Sigma methodology, DMAIC delivers sustained defect-free performance and highly competitive quality costs over the long run.

Step 1 – Define

The “D” (Define) in the DMAIC process focuses on selecting high-impact projects and understanding which underlying metric(s) will reflect project success.  In some cases the project metric will be a subset of a higher level business metric, as in the example below where warranty rates for a specific product line are being addressed.

In other cases the leadership team may already know that poor performance on a particular CTQ needs to be addressed, and achieving Six Sigma process capability on that CTQ would be the team’s goal.  The deliverable for the Define phase is a team charter to be reviewed and updated with the leadership team:

DMAIC Step 2 – Measure

The “M” (Measure) in DMAIC is about documenting the current process, validating how it is measured, and assessing baseline performance.  Some of the important tools in this phase include trend charts, basic Pareto charts, process flowcharts, Gage R&R, and process capability measurement (sigma level, also referred to as process sigma).
Depending on the project scope, the team might hold off on the process flowchart and Gage R&R activities until primary focus areas are identified further into the project.  In some cases, a macro flowchart is useful in providing all team members with an initial, high-level view of the process –

Process Flowchart - DMAIC Project

Oftentimes the Measure phase provides early clues that will direct team toward top problem areas and solutions, as shown in the following trend and Pareto charts for our outdoor furniture company example –

DMAIC Project - Trend and Pareto Charts

In the case of our furniture manufacturing company example, the project team also performs a Gage R & R study on a known critical measurement (screw torque) that affects a cracking condition on one of its chair lines – this will come in handy later in the project.  The initial GR&R results show that the screw torque measurement method introduces far too much variation –

Poor GR&R results on screw torque measurement

Using a graphical technique learned in Gage R & R training, the team finds that the first operator in the study is recording consistently higher readings than the other two operators –

DMAIC Team GRR Graphical Analysis

Finally, after correcting operator #1’s measurement technique and updating the measurement procedure, the team conducts a follow-up Gage R & R study with much-improved results –

Follow-Up Gage R & R Study


Finally, the baseline sigma level for the overall defect rate is estimated  using a sigma conversion chart, providing a relative indicator of how close the current process is to delivering zero defects.   A Six Sigma process has a sigma level of six, and for all practical purposes is considered a defect-free process over the long run, provided that adequate controls are in place to maintain capability.  In the example above, an overall defect rate of 2.6% reflects a sigma level of 2.1.

DMAIC Step 3 – Analyze

The Analyze phase in DMAIC isolates the top causes behind the metric or CTQ that the team is tackling.  In most cases there will be no more than three causes that must be controlled in order to achieve success – if too many causes are identified, then the team has either not isolated the primary causes or the project goal is too ambitious to achieve success with a single project.  There are always exceptions, but speed and results are key ingredients to building Six Sigma momentum inside an organization, and projects should be sized to assure team success and project closure inside reasonable time limits.

The Analyze phase deploys a number of tools for collecting team input and conducting objective experiments to identify or confirm top causes.  The most commonly used of these are –

Not all tools are used on all projects.  Trained Black Belts understand the process improvement tools available in each phase, and know when to use them.

In our outdoor furniture manufacturing example, the team must understand the reasons behind two of the biggest Pareto items discovered in the Measure phase:  (1) returns on the “AD Chairs” product line and (2) returns on Direct-Ship products.  A detailed analysis via product tear-downs produces the following sub-Pareto charts, which bring the team closer the the primary causes that are driving high customer return rates –

 Sub-Pareto Charts

There is still a great deal of additional work for the team as they proceed down the path to their primary causes in this case, but the sub-Pareto charts noted above will further narrow the team’s focus to (1) cracked back-slats, (2) shipping damage, and (3) missing hardware.

DMAIC Step 4 – Improve

The Improve phase focuses on fully understanding the top causes identified in the Analyze phase, with the intent of either controlling or eliminating those causes to achieve breakthrough performance.  The overall theme for the Improve phase is process redesign, and the following Six Sigma tools are commonly used in this phase –

The Improve phase also includes a limited production trial (or beta test for non-manufacturing projects) that demonstrates a significantly improved process-sigma-level prior to moving into the Control phase.

DMAIC Step 5- Control

DMAIC’s Control phase is about sustaining the changes made in the Improve phase.  The best controls are those that require no monitoring (irreversible product or process design changes).

But oftentimes there are process settings, setup procedures, etc., requiring that employees follow specific requirements in daily operations – these items are typically documented in a control plan.  In cases like this the Six Sigma team should do everything possible to error-proof the process, and should then add the appropriate checks and balances to the quality system for the long run.