The Jack factor

Then there’s what I like to call the Jack factor—based on an engineer I once worked with named Jack (not his real name). Jack didn’t want anyone to know jack (pun duly intended). He was the closest thing I have seen to a covert operative working in his own job. He didn’t want you to know him, or what he did and was very resistant to training anyone. When I questioned Jack about his behavior he proudly replied “It’s my strategy for job security. If no one else knows how to do my job you’ll never be able to replace me with a less expensive person.” The last time I checked, Jack was still in his dead end job doing remarkably well at holding back his own career as well as the ability of his organization to grow.

In order for the development of internal leaders to be effective in an organization:

1. The commitment to the program needs to be firmly rooted in the corporate culture and endorsed and acted upon by the top leadership.
2. Establish a plan with milestones and select a passionate champion to lead the effort (similar to the implementation of a quality system).
3. Authorize time, money, and resources and make it part of the performance evaluation process.

Does the development of internal leaders create more work, stress, and discomfort for the incumbents responsible for developing their staff? Most likely yes.

Does it raise expectations for additional responsibilities and higher salaries for the individuals being developed? Most definitely.

It is a challenging process. If this was easy we’d be better at it.

James Harden of the Oklahoma Thunder received this year’s sixth man of the year award from the NBA. Here’s to your sixth men and women.

Paul Billimoria
President
HRG, Inc.

I, like most others, have been guilty of this polarized way of thinking many times. One day I realized how limiting this approach is, and that there are a multitude of shades of grey in between.

My ‘aha’ moment of realization, where for a brief instant I felt as if I had come close to enlightenment after sitting under a Bodhi tree, came when I used predictive analytics for the first time. Up until that point most of the statistical analysis I had done involved studying correlation values among many variables, fairly basic by most standards. All my education on the subject to that point had been about analyzing the recent past, and making a guess as to how that might affect things in the near future, i.e. a moving average forecast. It’s a very simple method that is not very accurate, yet it is still widely used by many businesses today. Most businesses engage in this type of rear view driving when looking to the future, with good reason. You need to look at the past because what happened there will often set the trajectory for the future. But it doesn’t go far enough.

The first predictive model I created was designed to forecast unit sales of a retail product based on economic and demographic data. I had my equations, dependent variables, historical company data and I was all set to go fortune telling. As I finished making the model, I compared the sales forecasts that I had generated against the company’s actual sales and saw that almost all of them were within a few percentage points, save for a couple of outliers.

The model predicted 90% of the variability in the retail sales variable a year forward, while the historical trending model had an absolute margin of error nearing 40% within a few months of current.

I realized that my model and my predictions, while not perfect, were able to deliver insightful and accurate forecasts to the company, much more so than their traditional method of a moving average forecast that left much to be desired due to its large built-in margin of error. Even the couple of forecasts that were off by a lot ended up being just as insightful to the folks in charge of the company. They were able to identify why their sales deviated so greatly during those years.

Why were the inferential statistics so much more valuable than the trending data?

In brief, a moving average forecast assumes a linear relationship between time and the change in the value you are interested in. Once you reach the present, the line going through the past data is projected forward, with a hefty positive and negative margin of error. This method of projecting forward by looking to the past is a proverbial flip of a coin, because there is as yet no knowledge of why the trend was as it was. Businesses and company leaders can do far better.

With predictive modeling we find the relationship between the dependent variable (the metric the company wishes to predict) and the independent variables from prior years which may be leading indicators.

If you can identify the correlated variables, the result is a statistical equation where you can plug in data from one year (for example) and get a prediction for subsequent years, with much greater accuracy because the variables actually co-vary together and you don’t rely on the assumptions built into a linear analysis. This is an anti-coin flip because there is so much more information gleaned from the analysis of past data. It’s like seeing an old time TV show in Technicolor with special effects.

In these volatile, turbulent, and uncertain economic times, where yesterday’s gains are quickly mitigated by today’s losses, history is not the guide it once was. The old method of looking back to see where you are going is being replaced by predictive analytics.

Companies can settle for the flip of a coin that leaves a lot of insight unused, or they can choose a method that looks at all the possible relationships for a better and more in-depth understanding of the factors that drive their important decisions. Business analytics can be used to measure, evaluate, understand, and predict almost anything, so why flip that coin?

Evan Loker, MBA
Director of Analytics
HRG, Inc.