Six Sigma Revisited: Statistics, Process and Scale

Back in January we ran a story on Aaron Ripley, the director of quality and strategy at Cox Communications who was named Young Engineer of the Year at this year’s SCTE Conference on Emerging Technologies.

Ripley was commended in part for his knowledge and use of Six Sigma, a quality control technique with roots in the manufacturing sector, Motorola in particular. Its migration into the service industries, including cable, even as it faces criticism, is an intriguing story.

A note in my latest alumni magazine, of all places, mentioned that Cox VP of Operations Tom Guthrie had delivered a keynote address at the 2006 Institute of Industrial Engineers Quality and Lean Conference on "Practical Application of Six Sigma in an Information Technology Group" prompted me to give him a call and review some notes.

As that keynote title suggests, Guthrie works on Cox’s IT side. The Young Engineer honoree Ripley is in corporate engineering. Both arrived at Cox about five years ago, Ripley from General Electric (famous for its adoption of Six Sigma under Jack Welch) and Guthrie from some 20 years in systems engineering and IT management and with degrees from the US Military Academy at West Point and U.VA (go Hoos!)

Differences of background aside, the two appear to have a common outlook when it comes to the applicability of statistical tools. "I think Aaron and I have a very similar approach," Guthrie says. Objections overruled With its classic emphasis on very few defects per million opportunities (DPMO), Six Sigma has come under criticism for being narrowly applicable to existing and highly scaled (vs. innovative or disruptive) processes, arbitrarily based on six (why not five? Or four?) standard deviations, and something of an expensive and impractical fad.

Let’s look at those objections. As for the scale implicit in the notion of "millions of opportunities," the fact is: Cable operators today are highly scaled operations.

Guthrie says that metric is relevant in the voice business, with many millions of minutes passing over cable networks. As for defects, dial tone is such an assumption for customers that its absence is one among many possible areas to track.

"We’ve also had to take that approach with ‘data tone,’ if you like," Guthrie adds.

As for why six standard deviations, Guthrie emphasizes the need to choose both the correct ends and correct means. "It’s not just Six Sigma for Six Sigma’s sake," he says.

It’s a good practice for customer service, for instance, to answer the phone more quickly than the internal IT team, he suggests. The IT team might then aim for three rather than six standard deviations from the norm, making it (in Six-Sigma jargon) "capable," allowing a more forgiving 67 defects per thousand, rather than 3.4 per million.

"The cost difference between the two is quite profound," Guthrie says.

That attention to cost helps answer the last objection, i.e. that Six Sigma is an expensive fad. Guthrie emphasizes that it’s much more than an academic exercise. "It wouldn’t survive at Cox if it was," he says.

At its practical best, Six Sigma encourages the sound use of statistics. "I wanted a method that looked at data over time," Guthrie says. "Is it stable from a statistical standpoint?"

And here the cable industry’s traditional propensity toward tinkering (at least among engineers and technicians) and aversion to strict change management comes into play. If the process changes, then the data can lead toward apples-to-oranges comparisons; and if the process changes for the better, then you can end missing a sustainable improvement.

"Managers get focused on current data points," Guthrie says. "(This) really forces you to look at data in a time series." GE influence at Comcast? Cox isn’t the only one that uses sophisticated operational metrics, as anyone familiar with Comcast tech ops knows.

In our interview with Comcast EVP, National Engineering & Technical Operations, John Schanz for the 2006 issue of CT’s Communications Executive, for instance, we asked about his earlier experience at General Electric Information Services, about whether GE’s emphasis on scale had its limits and whether he saw himself as an engineer or business leader.

"Is scale a good thing? Yes, because you get greater economics and greater leverage across your assets. But you can never just do scale. You have to look at things within the scale to find innovation and emerging capabilities for the customer," Schanz said. – Jonathan Tombes