Can these methodologies be pushed up, way up, the product development process? Into the Research organization, for example? This article discusses some of the issues, and is for research and technology development professionals, as well as their managers and those responsible for Six Sigma and other improvement initiatives within their organizations.

If we apply Lean methods to research projects what will happen? Might we "lean out" our freedom to explore by establishing a bureaucracy of standard work? How can one do a capability study? There is no proven science yet, let alone customer requirements for some yet-to-be-defined product! The researcher's data is not normally distributed and pretty much everything they work with is highly non-linear, so typical Six Sigma course materials would be irrelevant. Plus, sample sizes range between 2 or 3 (if the researcher is lucky)! Agile principles were conceived for software developers (see http://agilemanifesto.org/); what's that got to do with the discovery of science? The only software code researchers might create are one-off applications for their personal use during data analysis. Lean, Six Sigma and Agile? Their typical application seems incongruent with this environment!

Here is an imaginary, but highly plausible, dialog between a scientist and a business manager:

Scientist [to Manager]: "I will let you all know when something I chose to do adds value! Until then, be patient and let me do my work."

Manager: "Fair enough, what are the things you plan on trying as a matter of course?"

Scientist: "Well if I had to write it down, I guess it would include forming a hypothesis about the ideas I have in my mind, then I would construct some mathematical concepts and perhaps graphical sketches that help me coax out the math I need to express myself. Then, when I get far enough, I usually create a physical model to validate my math model. Sometimes, I have to invent a way to measure the phenomena I am generating. So once I can measure the thing with confidence, I move on to generating empirical sets of data. These can be looked at for repeatability and stability. If I get that under control, then I can substantiate the terms in my math model - refine and tune my model. Once that is in hand, I normally see if I can change the terms in the model to define where the range of function stops being measurable and stable - where the science breaks down. If the dynamic range of the model is stable and repeatable - I've got my model and viola, a new bit of science is born."

Manager: "So, do you do that set of steps over and over every time you do your work?"

Scientist: "Yes, it's a pretty common set of steps but what goes on inside of each one can vary wildly depending on what I discover or run up against."

Manager: "So your steps are pretty consistent but you have a great deal of change and surprise inside any one of them?"

Scientist: "Yes, now you are beginning to understand my world! I live in an exploratory environment where I actually am looking for surprises and variations that I did not expect. To me variation is sometimes my best friend and at other times my worst enemy."

Three key things in this discussion should be evident to a person who is familiar with Lean, Six Sigma and Agile development methods:

1. It is easy for a research team to tell you what their repeatable and variable forms of work are. They can tell you what adds value and when, in the past, that has been the case more often than not. Most research scientists carry a set of lessons learned that form a set of rules they follow (heuristics) as they repeat their processes. They have an innate ability to define a "lean" flow of value-added work that is justifiable on a repeatable basis.
2. Scientists can go on for hours about the difficulties they confront every day in terms of measuring the phenomenon they are creating. Just getting a clear signal from their measurement of phenomena is difficult, let alone characterizing the noise associated with the capture and processing of that signal. Measurement system invention is often just as important as the invention of the new science they are working on. So they frequently have to invent twice for the sake of data integrity so people will believe their science. Six Sigma measurement capability is logical and, in fact, required to underwrite the integrity of the research data. The science is not Six Sigma - the measurement systems have that requirement at a minimum.
3. The abrupt interruption of planned scientific work, as scientists and researchers try to follow the scientific method, leads to new, unimagined avenues of exploration. Scientists often hit a wall as they do their work and have to turn-on-a-dime to make progress or confirm their hunches are dead ends. Flexibility and latitude are needed to explore unanticipated discoveries. Project planning involves the negotiation of these unpredictable tasks and how many branches can be explored before funding runs dry. Agility in project planning and adaptive freedom to explore by design is essential to discovery. The funding and resources are not unlimited, so a negotiated amount of freedom is the core of the risk management of a research project. Agile principles can help with this.

The original 13 or so tools of Classic Six Sigma have grown considerably when work is focused on preventing problems - rather than reacting to them. Certainly design and manufacturing engineering professionals have proven that Six Sigma works well in their processes in the form of Design for Six Sigma (DFSS). DFSS has become an accepted approach to enhance the ability to commercialize new products and the Six Sigma DMAIC process has been used to improve existing designs and the processes that make them.

To say there is an identical paradigm in a research organization is a stretch. The scientific method is as close as we can get to a historical foundation for standard work in a research laboratory. It is likely (and we have observed) that companies attempting to teach DFSS to research groups have largely met with failure and cultural rejection. Taking standard DFSS up stream and dropping it into R&TD is not recommended. Strategically adapted according to what scientists say they can add to research processes and projects, however, Lean, Six Sigma and Agile methodologies can be highly successful.