Spurred by Jakob’s latest column on Six Sigma and usability, and my own experience with financial services organizations trying to introduce Six Sigma methods into their design processes, I asked the CHI-Web mailing list if Six Sigma applies to user experience work.
There have been some great responses (more informative than Jakob’s original essay). I call your attention to two:
Julie Jensen talking about utilizing Six Sigma in the usability processes at USAA.
Robin Jeffries discussing how it really does enable smart user centered design (at Sun, at least.)
This is great stuff, and I hope to hear more!
(I originally thought of titling this post something like “I don’t know if I like all this discussion around “Sex Smegma”, oh wait, it’s “Six Sigma”? Never mind” in some Emily Litella-like fashion, but that seemed too confusing.)
Here’s a question: Do companies want to pay for getting their users to only 3.4 failed tasks per million tries (coopting the 3.4 defects per million noted in Jakob’s article)?
If so, no problem. Let’s test 1437 people, and we will need an extra 8 months to do it, and by the way, how about all those resources we will be tying up…
As Julie noted in her post: The need for the business to have design recommendations and solutions out ranked the need to document Six Sigma statistics, especially at the cost of 200 test participants!
It’s nice to see that recent buzzword of ROI hasn’t left the minds of business people with the addition of dashboards, quality assurance, customer experience, and risk management.
On a final note, isn’t what Jakob talks about DMAIC in usability terms exactly what UCD is? I guess I don’t get how this is a new thing.
The new part I suppose is the defect tolerance level of six-sigma, but I would bet that most companies don’t really care that much. And if they do, please hire me!:)
I’m not sure how your Six Sigma folks determined you needed to test 200+ users. I have found that these sampling problems stem from a mis-categorization of what’s being tested. See my articles on usability sampling:
Do you have their calculations? Is it for binary data (Task completion) or continuous and what confidence level are they testing with? 95%
You don’t need hundreds of users in order to show statistically significant results or use six sigma.