top of page

Tom Cruise, Chocolate Chip Cookies and Six Sigma

Minority Report is a science fiction movie starring Tom Cruise and directed by Steven Spielberg. It takes place in the year 2054, where PreCrime, a specialized police department, arrests criminals before they commit crimes. It’s the job of psychics, called ‘precogs,’ to identify these would-be crimes.

Want your own precog?

What could you use a precog for in your organization?

  1. Identify which employees are at most risk for leaving, so you can take action to retain them.

  2. Determine how likely it is that a new customer is going to pay your invoices on time, to assist you in granting credit.

  3. Decide which potential location for your next retail store will result in the highest revenue.

Well, if you don’t know any precogs or you’re not as comfortable as the future government in enslaving the ones you do know, perhaps your HR leader (in example 1, above), Credit Manager (example 2) and VP of Marketing (example 3) should investigate Linear Regression, a statistical tool often employed by Six Sigma practitioners.

A key concept of Six Sigma (6𝞂) is that an outcome or result is related to one or more inputs. To improve the result, you need to look at those inputs.

Let’s say your family is not happy with the chocolate chip cookies you’ve just baked. In 6S terms, the desired output is a delicious and attractive cookie. What are the inputs?

  • Properly measured, high quality ingredients

  • Temperature of the oven

  • Time of baking

  • Proper greasing of the pan

  • Knowledge of the features of the cookie

So, a tasty cookie (Y) is a function of those input factors (X’s). Or,

Y= f(X1 + X2 + X3 ...)

A linear regression would help you investigate, based on all your previous batches of cookies, how each of those input factors is correlated with the output (a tasty cookie). Assuming you can numerically rate the resultant cookies, this technique will come up with a formula that will specify the ideal settings for each input factor to result in the best possible cookies.

LIkewise, you can input known characteristics of former and current employees to determine who’s at most risk to leave.

I used to work with a Senior Vice President of Human Resources, a former Six Sigma Master Black Belt. In a previous company, he inherited a situation with more than 100% turnover of employees annually. In other words, with an employee base of 500 people, his department had to hire at least 500 people each year to remain fully staffed. Never mind the impossibility of finding that many qualified people each year, just think of the cost - some organizations put the cost of turnover at 30% of the salary and benefits of the employee.

He reviewed employee records going back two years, both those who left and those who were still with the company. With that as the input for his linear regression, he determined an equation to use to determine after how many years would an employee leave the company, voluntarily or otherwise. The answer (Y) ranged between 0 (the employee would not make it to their first anniversary) to 47 (an 18 year old would stay until retirement at age 65). Factors considered included age, marital status, years with the company, years in position, commuting distance to work, current salary, time since last pay increase and more.

He obviously had to keep the data confidential. He also made sure that only positive actions were taken (no pre-emptively taking retribution on an employee identified as likely to leave.) In line with published studies, he found that compensation-related factors were not highly correlated to the propensity to leave.

To put the new information into play, after running each current employee’s data through the equation, he identified who was predicted to leave in the next 6 months, repeating the exercise every 6 months. For anyone predicted to leave within that period, he’d meet with the employee’s manager to determine strategies to retain the employee. He also worked with the leadership team as a whole to address common themes around employee dissatisfaction derived from the correlation study.

This exercise moved the conversation from a typical annual employee satisfaction survey (why would you be happy or unhappy at work) to an analysis of what actually happened (why WERE you happy or unhappy, which led to the decision to leave or not). This is analogous to the reason I’m not a big fan of Voice of the Customer (VOC) measurement. VOC asks a customer how likely they are to do business again or to recommend the company to others. That’s like an election poll. Saying what you’ll do is not the same as observing what you actually do.

The result?

Within that first year, the turnover rate was reduced by around 80%. He attributes the tremendous turnaround to opening the eyes of his fellow executives as to how to make the workplace more appealing in general, as well as to the attention granted to individual employees based on identifying them as high-flight-risk.

Still think that Six Sigma is only applicable to manufacturing? Linear regression and the rest of the gamut of 6𝞂 tools and methods offer equal improvement opportunities far away from the production floor.

Or do you want to continue serving your family burnt, tasteless chocolate chip cookies?

by Dave Boss, Managing Consultant, The Operations Group, LLC, ©️ 2020

bottom of page