Teaching a database to lie

What do numbers mean if we change them to say whatever we want them to say? At my previous job, I was given an interesting assignment -- something quite different from my usual IT chores. My boss asked me to create a way to measure the performance levels of our salespeople. Management wanted an objective way to see whether people were selling as much stuff as they were supposed to be selling. It seemed simple en

What do numbers mean if we change them to say whatever we want them to say?

At my previous job, I was given an interesting assignment -- something quite different from my usual IT chores. My boss asked me to create a way to measure the performance levels of our salespeople. Management wanted an objective way to see whether people were selling as much stuff as they were supposed to be selling.

It seemed simple enough. I built an engine in Visual Basic; then I created a database and spent a lot of time entering fake records and comparing them to an Excel spreadsheet to ensure everything was calculating correctly.

Nonetheless, after the app went operational, I started hearing complaints from Sales about how the numbers were lower than they should be. My thought was that the numbers might be low because people weren't hitting their goals. Hmmmm ... . Anyway, at the insistence of the VP of Sales, I changed the formulas to account for sick leave, vacation time, holidays, and official leave. I'm surprised he didn't order me to include restroom and smoke breaks.

Even so, the numbers still weren't right -- not, at least, according to Sales.

Of course all of this was being discussed in management meetings (to which I was never invited) with the Sales people insisting that my app was inaccurate, that we couldn't use these results. Eventually someone told me they were looking for an outside consultant, since it appeared that I wasn't able to complete the project successfully. I said that I would be happy to discuss any concerns about my apparent inabilities. (Well, I didn't say it exactly like that.) I think someone figured if we paid a consultant enough money, he could make the numbers say things that weren't true.

Finally I was called into a meeting to discuss my "inaccurate" numbers. Before the meeting, I went through the database and used an Excel spreadsheet to prove that 2 plus 2 really did equal 4. I even killed a few trees and wrote out manual calculations so people could have a crystal-clear picture of how the database worked.

I began by going over the formula ... again. Next thing you know, my biggest opponent in Sales says, "That's not the issue. We know the formulas are correct."

Huh? Inaccuracy was all I ever heard from every direction, which is why I had wasted my time writing out the calculations. Since it was now apparently agreed that accuracy wasn't the problem, I asked what the problem might be. The head of Sales declared that they were convinced that my database set the expectations for meeting sales goals too high.

They never explained how this was my fault, especially since all the underlying assumptions (and expectations) had come directly from management.

But Sales had a great solution. Say, for example, someone had a sales goal of 65 percent, and week after week they were only doing 60 percent. My job was to rewrite the data entry screen so the tech typing in the data could change their expectation from 65 percent to 60. Sweet!

I almost asked, "What's the point of measuring sales performance if whatever they sell is acceptable? What do numbers mean if we change them to say whatever we want them to say?" But I knew better than to waste my breath. I'll spare you the remaining details of this nightmare. Happily, I'm now in a new position where decisions are made with at least an occasional nod to logic.

From CIO: 8 Free Online Courses to Grow Your Tech Skills
Join the discussion
Be the first to comment on this article. Our Commenting Policies