Intelligent Enterprise

Better Insight for Business Decisions

RSS
Webcasts
Whitepapers
Subscribe
Home

June 17, 2003

BI's Promised Land, Part II

Measuring processes is an important step, but improving processes is critical if you want to realize full performance management benefits

by Erik Thomsen

Continued from Page 1

So what are the drivers of customer satisfaction? A friendly or knowledgeable salesperson, a generous return policy, low prices, occasional steep discounts, and store location or hours? Assuming that these are the drivers, how are they related? How does lowering average prices affect the amount of customer satisfaction improvement generated by increasing the generosity of the store's return policy?

Representing customer satisfaction as the outcome of a set of drivers, such as the ones listed, is tantamount to writing a classical regression-style equation. In customer satisfaction, the dependent variable is a function of some number of independent driver variables:

Customer satisfaction = f ( D1, D2, D3, ...).

The better you know the determinants of the outcome (in this example, customer satisfaction) the more you can improve the quality of that outcome by avoiding errors in the processes that contribute to it. Taking the preventive approach saves you from spending energy producing faulty outcomes, such as unsatisfied customers, and then spending even more energy correcting them, such as offering special discounts to win back disgruntled customers).

As obvious as this advice might seem — and as much as practitioners of statistical process control and Six Sigma have stressed it — business at large hasn't embraced this approach. Companies still test the quality of manufactured products (from clothes to cars) rather than the quality of the manufacturing process. Vendors still systematically test software after it's created; bug fixing and patch creation consume non-trivial amounts of energy. And companies spend a lot of time appeasing unhappy customers.

If it's so obvious, why don't businesses simply measure and improve drivers? The answer, of course, is that the drivers of process outcomes aren't necessarily obvious. You have to acquire the knowledge of those process drivers, which lets you improve those processes.

Tuning Out To Tune In

How do you learn about the drivers of an organizational process? Let's say you're the chief marketing officer for a large holding corporation that owns five separate retail businesses, each with its own URL and Web-based sales. You're trying to learn from the experiences of the separate businesses what kind of Web site layout works best. Looking through the various URLs, you've noticed that some sites employ a wide and shallow site layout while others employ a narrow and deep layout. Each layout has its pros and cons. Discussions with the marketing heads of the distinct units reveal that, over the years, each business unit switched layouts; that is, those that currently use a wide and shallow layout at one time used a deep and narrow layout and vice versa. Between the five units, you've got more than 4TB of historical Web activity data collected over a period of eight years.

			>