To Model or Not to Model, That is the Question


As a senior analyst for Business Development, I create models that closely resemble real problems. True to what they represent, models come in all shapes, sizes, and styles. Some models are intangible, some tangible. Ultimately, they’re human constructs used to better understand a real world construct.

No matter what kind of model you are building, here are the basic steps you need to follow:

1) Define the business problem. Communication is the biggest business problem today. Einstein is quoted saying that if he had one hour to save the world he would spend “fifty-five minutes defining the problem and only five minutes finding the solution.” This quote illustrates that before jumping into solving, one should fully understand the problem. Not clearly understanding the business problem will result in several model revisions or a useless model.

2) Clearly define the time period. Business models typically vary over time. For example, a retail sales model should likely show higher sales during holiday periods. A new inside sales team sales model will generally require rep ramp time. A partner recruitment model will generally require partner ramp time. Therefore, time periods must be considered in model to accurately reflect the real world.

3) Determine the output variables. Lewis Carroll, author of “Alice in Wonderland,” wrote “If you don’t know where you are going, any road will get you there.” This saying applies to modeling. If you do not know what you want your model to show, there is no reason to build it.

4) Determine the input variables. Inputs are the variables being manipulated or changed. One of the greatest challenging of modeling is determining the important input variables. Too many inputs can create an unmanageable model. Too few inputs can create an unrealistic model. From a modeling perspective, incorporating only the important variables into a model provides a simpler, more useful, and more reliable model; the model will also be more practical to apply because fewer variables need to be measured.

5) Test the model. There are many instances when you do not know if the model outputs are realistic. To validate the model, do some research on similar proven models and determine if your model is providing similar results.

When you’ve got the process down, onsider the following questions models are designed to answer:

• What marketing campaign will generate optimal revenue?
• Where geographical areas should I market to?
• How many reps do I need to achieve optimal coverage?
• What stores should I market to?
• What is my return on investment?
• What is my cost of sale?

Please stay tuned for my upcoming posts. I’ll be sharing more examples of models we’ve used at MarketStar to solve real business problems. How do you use models? Feel free to share in the comments.

About Phil Rogers

Phil Rogers is a veteran statistician , with more than 15 years' experience as Lead Analyst for several company contract teams managed by MarketStar Corporation. He configures databases to capture the necessary information for reporting, database managing, and creating views using Microsoft SQL. He develops weekly, monthly, and quarterly standard reports, creating reports using various tools Microsoft Excel, Microsoft PowerPoint, Microsoft Word, Qlikview, Microsoft Access, SAS, JMP, Statistical forecasting models and Statistical ROI models. Rogers teaches business statistics and calculus as an adjunct faculty at Weber State University.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>