DMEDI or DMAIC? That is the Question.
As the acceptance of Six Sigma has grown and penetrates deeper into common business
culture, a question has begun to arise to determine when does a process need to be improved
versus designed? How do you decide which roadmap is better for a given problem, DMEDI or
While clear to veteran practitioners, this question is not quite as clear to the general business
public or newer Six Sigma professionals. This question also exists primarily due to the way Six
Sigma has been sold to the business community. Most organizations have been sold on Six
Sigma DMAIC being a “quick fix” to internal organizational issues and problems. However, the
true benefit Six Sigma brings comes when the maturity model moves from fixing broken
processes to properly building processes from the onset.
This benefit will only be seen over time. But in the meantime Six Sigma professionals will be
faced with receiving project charters and needing to decide how the problem should best be
Before examining the decision process to properly select the best roadmap for a given project,
we must review each methodology for what it is good for and what it is not good for. Here is a
definition and breakdown of components and steps for each roadmap.
DMAIC (Define, Measure, Analyze, Improve, Control)
DMAIC is an analytical, data driven approach to eliminate weaknesses in active processes,
products and services. DMAIC brings incremental improvements.
DMEDI (Design, Measure, Explore, Develop, Implement)
DMEDI is more of a creative approach to designing new robust processes, products and
services. This roadmap is focused purposed to obtain significant competitive advantages or
quantum leaps over current environments. However DMEDI Projects tend to be more time
and resource intensive.
Let’s briefly compare and contrast each roadmap step by step. Essentially the Define phase is
the same for each methodology. Both Define phases are purposed to provide completed
charters with clearly stated business problems, desired results and scope limitations. Since
there is little differentiation between the two, we can move on to the second step of each phase.
The Measure Phase
DMEDI and DMAIC are similar in that they have five phase steps to completion. After the first
common Define step the methodologies begin to differ significantly. The most important
differentiation in the process lies in the second step, Measure. At first glance the steps seem
the same for both, but under closer examination we can see that the measure phase in the
DMEDI methodology is significantly more involved. The measure phase in DMEDI requires
more examination than DMAIC because there is little if any existing process definition, baseline
outputs (Y’s), or critical customer requirements (CCR’s).
Because of the lack of existing definition, the need for CCR development is much more
significant in the DMEDI Measure phase. To meet the need for clearly defined CCR’s the QFD
or house of quality is used in multiple iterations and phases. In the DMEDI path the house of
DMEDI or DMAIC? That is the Question – Steven H. Jones January 2006 Page 1 of 4
quality can actually become a village. This can prove necessary to properly define a product or
service that the customer truly desires.
Analyze versus Explore
While these two steps are similar they have a significantly different deliverable. The DMEDI
Explore phase is purposed to deliver a conceptual design of a new process. The Analyze
phase of the DMAIC process is purposed to breakdown existing data of an existing process to
identify potential root causes. Here the DMEDI phase is conceptual and the DMAIC phase is
Improve versus Develop
Here again we have two similar phases that also have differing deliverables. They are similar in
that they both are both purposed to deliver a new process. Specifically, in DMAIC the Improve
phase is purposed to produce a rational future state design and DMEDI is purposed to delivery
an optimal design. One difference is that the rational design is based on statistical or
mathematic proof and the optimal design is based largely on meeting customers’ desires.
Additionally, the phases differ as the DMAIC Improve phase should contain a live pilot of the
revised process, whereas the DMEDI Develop phase process lives only on paper.
Control versus Implement
The key differentiator between these two phases lies in the piloting of a new process. In
DMAIC, a temporary, small scale future state pilot has already been conducted in the Improve
phase. In DMEDI the pilot is conducted here in the Implement phase. Both pilots seek to
validate the capability of the proposed process to meet or exceed the project objectives and
identify problems. But the DMEDI Implement phase pilot is normally a permanent, full scale
deployment unlike the small scale pilot in the DMAIC improve phase.
Beyond the differentiation in the location and scale of the process pilots, the balance of the
functions of the Control and Implementation phases are quite similar. Both are purposed to
deliver comprehensive control plans and charts to monitor the activity of the new process.
The greatest difference between the two methodologies lies in their unique purposes.
The DMAIC methodology’s resources are spent to reduce the problem, be it waste, cost
or time. The DMEDI methodology’s resources are spent to prevent potential losses.
Here is a side by side examination of DMAIC and DMEDI Six Sigma roadmaps
Define – Determine Project Scope, Define – Very Similar to DMAIC
Objectives, Resources, Constraints
Measure – Determine Customer Groups, Measure – Define Customers and Needs
Determine CCR’s, Obtain Data to Quantify Using VOC and QFD, Determine CCR’s
Analyze – Analyze Data to Identify Tangible Explore – Develop Design Concepts, and
Root Causes of Defects High-Level Design
Improve – Intervene in the Process to Develop – Develop and Optimize Detailed
Improve Performance Design
Control – Implement a Control System to Implement – Validate Design with Pilot,
Maintain Performance over Time Establish Controls, Full Scale
DMEDI or DMAIC? That is the Question – Steven H. Jones January 2006 Page 2 of 4
Now that we have examined a comparison of the two methodologies and contrasted their
differences, let’s look at each through real world environments to see how each model can
Reduce cycle time and errors on Design of new project
service orders management office
Increase first call resolution on Develop a new service order
support calls handling process
Service order cycle time reduction Develop a new contract renewal
Product manufacturing/assembly Manufacturing process requires
cycle time reduction major redesign (conversion from
IT solutions – improve back up gasoline to alcohol engines)
and recovery time, reduce Welding new materials
patching time Design of new triage process for
Identify causes and eliminate new technologies
defects in molds Develop new formats of data
Eliminate false server alerts storing larger files
Reduce product development New fuel injection design
cycle time New material development
Identify causes of defects in a Next generation ink delivery
copier design system
Reduce defects released in new
Since our original question was how to decide when to use each methodology let’s examine the
decision inputs. Properly deciding whether your project is going to be a DMAIC improvement
project or a DMEDI design project will be essential for real success.
There are two key questions that have to be asked in order to properly decide which
methodology will best suit a given problem. They are, “Does any process currently exist?” and
“Is a wholesale improvement necessary?”
If the process does not currently exist in any format at all, then DMEDI is clearly the way to go.
But this is a tricky question. In many organizations, processes completely lack documentation,
but the work is still being performed. The process may even be performed differently by
different teams, but it does exist. When a process does in fact exist, even in a very loose form
the DMAIC roadmap should be followed into the Measure phase to set a statistical baseline to
In some cases a process may currently exist but be so fragmented or badly broken that to reach
the objective it is best to start from scratch. In these cases where the desired objective and the
current performance are light years apart, the need for a quantum leap in performance may be
chosen. This decision will need to be made after examining the baseline data and objective of
the charter. The main reason for not automatically opting for DMEDI is time. DMEDI projects
typically require a longer lead and resource time to complete. You won’t see too many Kaizens
in DMEDI projects.
DMEDI or DMAIC? That is the Question – Steven H. Jones January 2006 Page 3 of 4
Now that we have reviewed all of the components of each a discussed the thought process
around selecting the most appropriate one here is a graphical decision tree to assist with the
No Y es
DMEDI Measure Measure
Design new Improve
product(s) or existing
Im p le m e n t
No gradual Y es
Steven H. Jones is a Process Engineer who received his certification as a Lean Six Sigma Black
Belt by the George Group while employed and Xerox Global Services. He started his career at
the 3M Corporation, an early adopter of the Lean Six Sigma methodology in 1988 and has
worked in quality improvement of Telecommunications and IT arenas since 1993. Since then he
has provided quality improvement and process engineering services domestically and
internationally to clients such as BP Canada, Convergys, Intercontinental Hotels, and Microsoft.
He is currently a Senior Process Engineer with Siemens Business Services and can be reached at
DMEDI or DMAIC? That is the Question – Steven H. Jones January 2006 Page 4 of 4