Any undergraduate economics student will tell you, “The past is not an accurate predictor of the future.”  It’s one of the basest mantras of the entire field of study.  Historical returns on dot-com stocks in the late 90’s gave no clear indication when the bubble might burst [1].  The steady decline in year-over-year GDP growth seen since 2009 in the US didn’t even point in the same direction as the modest bounce we experienced in 2010 [2]. Using past years’ data as one’s only indicator of future performance is clearly an ineffective management tool.

In a recent project for a high tech client, AGI built a statistical model to predict future variable sales compensation expenses based on, among many other factors, historical performance against quota by the global sales organization.  A thorough analysis of the sales force’s past data laid the foundation for a model which let us simulate this year’s performance under similar circumstances.  Not surprisingly, our simulations produced estimates in line with last year’s results.  Rather than serve as a portal into the future, a mathematical model built upon sound assumptions is well designed to do nothing more than “predict” past datasets [3].  Something was clearly missing.

Abandoning one’s reliance on data is not the answer; on the contrary, it’s an invaluable source of guidance for many reasons.

  • First- it’s objective.  It will not bend or sway in the face of trends, fads, gut instincts, etc.
  • Second- it’s abundant.  Our post on Big Data asserts that even small firms already have more data than they know what to do with.
  • Third- it’s verifiable.  All data has interesting stories to tell. Once elicited, analysts can test and re-test those stories ad infinitum.

Data is the key, in this discussion, to understanding the causal relationships underlying what it is one’s trying to model.  But to make more useful predictions about next year’s sales activities, one must strike a balance between what past data tells us happened last year, and what business leaders think may happen next year.

What was missing from our model was an ability to do “scenario testing”!  By combining the objective underpinnings of a data-based model with the ability to “slide the scale” on its key assumptions, sales leadership can see what’s likely to happen under different sets of circumstances.  With extra information about how “good” or “bad” things could be (and not simply what a black-box model says they’re going to be), sales managers can hope to avoid unexpected bumps down the road, along with the “knee-jerk” reactions that sometimes accompany them [4].

Learn more about how Alexander Group’s sales analytics can help your organization avoid unexpected sales bumps.

Originally published by Ben Koupal.