Six Sigma History and Overview

Six Sigma was started in manufacturing for processes that are duplicated thousands and millions of times in something like the placement of a spot weld on sheet metal on a power turbine and being able to predict the failure rate of that part’s weld. If you can measure and control the variability in the manufacturing of this metal, you can reduce your defects and make a more reliable product. 99% effective isn’t good enough.

Then GE and a few other companies took the same approach and applied it to non-manufacturing processes like hiring employees, call center resolution time and debugging software. The jump to software puts some limitations on the use of the Six Sigma methodology; in this case you’re ostensibly making one product not thousands. You can still measure the process by which the software is made and remove defects in that process. The state of current Six Sigma in software is almost exclusively focused on defects like software bugs and coding errors. Very little has been done in the area of user analysis and user interface improvements.

Tenets of 6σ

Six Sigma is based on a few simple notions:

  • Customers defining what is an acceptable level of quality
  • Identifying an unacceptable condition and calling it a defect
  • Identifying the total number of opportunities for a defect
  • Stating numerically the probability that a defect will occur—or if there are 1,000,000 opportunities, how many times does the defect occur.

Human Factors Engineering and Usability Engineering both share the fundamental tenet of knowing the user. Six Sigma simply puts some well tested quantitative measures and a comprehensive process around knowing your users and their needs.

Usability and UCD are not just Qualitative

Another major concern about using the methods of Six Sigma for the user-centered design process is that UCD is too qualitative to measure quantitatively, much less so with statistical significance. Human behavior is very measurable as shown through psychology and many common psychographic measures. As most practitioners know, designing an effective UI is more art than science. For that reason the need for artistic design talent for excellent design will never go away. However, merely because artistic (right-brained) inspirations are used doesn’t mean quantitative measures cannot be taken. Whether web-based or through desktop software, every UI has some measurable goals:

  • Managing finances
  • Creating a document
  • Purchasing a product
  • Providing information about a product or service
  • Informing via news or some article

Can you think of a website, web-application or software product that doesn’t have any goals (email me and let me know)? As long as there are goals and users, quantitative measurements are germane. Goals can usually be translated into tasks and tasks have multiple ways of being measured. In fact, it’s been well documented that a “sense of usability” comes from five major usability rubrics (Nielsen, Usability Engineering: 1993)

  1. Learnability
  2. Efficiency
  3. Memorability
  4. Errors
  5. Satisfaction

ISO (9241)defines usability a bit more broadly as:

  1. Effectiveness (task completion and error rates)
  2. Efficiency (task time and learning time)
  3. Satisfaction (attitude rating)

Both Nielsen’s and ISO’s attributes provide many ways to gather quantitative measures about ease of use. After all, if you can’t measure your product or process do you really know how effective it is?

For more information on the fundamentals of Six Sigma check out the tutorial sections at isixsigma. For a discussion on qualitative/quantitative data and objective/subjective data and some misconceptions see Usability Test Data at Userfocus.