Business Intelligence Network - The Vision for BI and Beyond

Printer-friendly

E-mail to friend

Mea Culpa: What Do You Do When Mistakes Happen?

Published: August 9, 2006

There are two basic reactions to broken processes: the defects can either be ignored or acknowledged.

The fact of the matter is that even those of us who consult in and teach information quality processes can have IQ problems. I had one in last quarter’s article, The Economics of Information Quality. Two incorrect dollar amounts appeared in Figure 3, which ironically depicted the costs of ownership of an application in which a process improvement initiative eliminated 90 percent of the defects and costs of an order fulfillment process. The incorrect amounts are in row 2 ($25,000) and row 3 ($2,500) as shown in Figure 1 (which was Figure 3 of original article).

Figure 1: Incorrect Table Showing Costs of Ownership of Improved Processes

Figure 2 shows the correct amounts: $250,000 in row 2 and $25,000 in row 3. The difference in the five-year cost of ownership is $1,237,500, with actual five-year savings of $11 million instead of the reported savings of $12 million over the cost of ownership of the original process with its 20 percent error rate.

Figure 2: Corrected Table Showing Costs of Ownership of Improved Processes

To view the corrected article in its entirety, please link to it at http://www.b-eye-network.com/view/2800.

Responses to Errors
There are two basic ways to react when we have broken processes. One, we can ignore the defect(s) produced and hope no one notices. Or, we can acknowledge the defects and seek to make positive change. Errors left uncorrected can be far more harmful in the long run, but not analyzing root cause with improvement to prevent recurrence is irresponsible.

Analyzing the Cause of Errors
The precipitating cause was that I picked up the amount from the wrong row (row 3) of Figure 2 for the row 2 amount in Figure 3 and calculated the post process improvement amount by multiplying 10 percent times $250,000, instead of the correct value of $2,500,000. The spreadsheet calculation cascaded the error to row 3.

However, knowing the precipitating cause does not help prevent recurrence of the errors. I do have a quality assurance process in which two people in my office besides myself read my draft articles. I had asked them to check the figures, but had failed to ask them to check the figures and calculations from the description of the improvement in the text. As presented in the formulas, the data in each spreadsheet calculated with the same result as when calculated with an independent calculator. But the data was wrong, as the result was wrong, and this method of quality assurance was faulty.

Another cause was in running late in writing the article itself. As a result, I did not carefully compare the data and notice the (rather obvious) error that I should have caught.

Analyzing the Costs of Errors
I have spent about four hours of my time in reviewing and correcting this problem. My staff have spent an additional three hours getting the data corrected in our files and in working with Business Intelligence Network staff to correct the original article that is now archived and available to you. I caused Business Intelligence Network staff time required to make the updates.

However, an effect that is most troubling to me is that you, readers who may have used this in your work to gain commitment to IQ, may have received a skeptical response in having an error in the data if you presented the figures in the article. I apologize to you strongly if this happened to you.

Process Improvements from This “Effect”
We have improved our process:

1. Our quality assurance process to review calculations now compares calculations not just to data in the spreadsheet but also text to the description of how the data should be calculated.
   
2. We have also changed the schedule in order to minimize the last-minute rush to publishing.

I Am Concerned
I must share a concern with you. Why did I only hear from one person about this error? Did you read the article? If so, did you just accept the data without analyzing it, since it came from what should be a “reliable source?” Or, did you see it and laugh that it came from a guru? Or, are we living in such a busy world, that we do not have time to read things critically?

Prevent Recurrences
Information quality is really about acknowledging and studying the problems we have, analyzing the root causes, and finding improvements that will prevent or significantly reduce the recurrence of defects that waste time, money, materials and facilities of both ourselves and our customers.

What do you think? Let me hear from you at Larry.English@infoimpact.com.