Two Sales Forecasting Courses

We offer two e-learning courses as our core curriculum. The first covers the fundamentals of sales forecasting. Everyone involved in sales forecasting should master these basics. To learn more about the course, click the button below.
Our second e-learning course takes your sales forecasting to a whole new level of accuracy. It is a powerful course, yet highly accessible to a wide audience. For more information, click the button below.
We also offer instructor-assisted "Plus" versions of these courses.  Plus courses feature instructor review of homework assignments applied to company-specific forecasting challenges, along with multiple coaching webinars. Further details can be found by clicking on the above buttons. 

Our Forecasting Philosophy

  1. No complex formulas. You don't need to read "Greek" to do high quality sales forecasting. Let your software do the heavy lifting.
  2. No statistical significance. Far too many inaccurate forecasts come from "statistically valid" models. We teach you how to measure forecast accuracy the right way—simply!
  3. Test first. Hold out some of your most recent data and make your forecast model prove its accuracy before putting company money at risk.
  4. Just the best, not the rest. There are scores of forecasting formulas, but only a few really good ones. We focus on the best and ignore the rest.
  5. Everyone can forecast. When you focus on the fundamentals, anyone can forecast, not just geeks and wizards.

Two Schools of Thought

Conventional Wisdom

  • Manual overrides of statistical forecasts are more accurate than pure statistical forecasts.

Our Approach

  • Manual overrides often reduce forecast accuracy.
  • If you skimp on forecasting fundamentals, no amount of expert opinion can save you.
  • The key to good forecasting is picking the right forecasting software and method.
  • Forecasters who focus on software and methods run a high risk of overfitting. Noise becomes signal, or random error is mistaken for causal relationship. The result: high forecast error.
  • If your forecasting method shows statistical significance with historical data, then you are ready to make a real-world forecast.
  • Forecast as if it's your life savings. Test forecast models before risking company profits.
  • Divide your time series into in and out-samples. Use the in-sample to predict the out-sample. If successful, then make a real-world forecast.
  • There is no best way to measure forecasting error.
  • Error scaling (called MASE, or mean absolute scaled error) is clearly superior. See Rob Hyndman article link below.
  • A performance standard can be created for any time series, which means value added can be quantified for every forecast.
  • Past error measurement techniques are ambiguous, overly simplistic or too complex. Error scaling is simple yet powerful.
  • As the axiom goes: you get what you measure. If you have a great forecast accuracy measure, then you can make great forecasts.
  • Data mining is the next best thing in forecasting.
  • Causal data from well designed market tests (e.g., running a promotion in one region and comparing results to a "control" region without the promotion) greatly enhance forecast accuracy.
  • The risks of data mining are foolishly disregarded. When millions of statistical relationships are used to predict a single sales variable, spurious findings become near certainties.
  • You can forecast anything; it's just a question of finding the right forecasting method.
  • Some things are not forecastable; they are far more random than predictable. This inconvenient truth can be detected. The consequences of treating something as predictable—when it isn't—can be quite severe.
  • Forecasters should report statistical significance.
  • Forecasters should handicap every forecast based on prior accuracy, or best and worst case scenarios for key assumptions.

Top 4 Reading List

Rob Hyndman
Another Look at Forecast Accuracy...
 (2006)
  • Read the first page callout box.
  • One of the most important forecasters of our time.
Nate Silver
The Signal and the Noise (2012)

  • Fun reading.
  • Delivers on the promise of explaining "why so many predictions fail — but some don't."
Nassim Nicholas Taleb
Antifragile: Things That Gain from Disorder (2012)
  • Exposes the myths underpinning conventional forecasting wisdom in today's culture.
Charles W. Chase
Demand-Driven Forecasting (2009)

  • Great overview of the forecasting discipline. 
  • Written from the perspective of a major software vendor (SAS)