The Business Forecasting Deal: Exposing Myths, Eliminating Bad Practices, Providing Practical Solutions

by Michael Gilliland

Practical-nontechnical-solutions to the problems of business forecasting Written in a nontechnical style, this book provides practical solutions to common business forecasting problems, showing you how to think about business forecasting in the context of uncertainty, randomness and process performance. Couched in the context of uncertainty, randomness, and process performance, this book offers new, innovative ideas for resolving your business forecasting problems.

Genres: Business

272 pages, Hardcover

First published May 11, 2010

Reviews

[Ben]

The tone felt off on this one. Rather than focus on the need for a mitigation strategy to counter the risk of inaccurate forecasts, the author repeatedly focuses on how the items to be forecast are not able to be forecast correctly. This gave the book a decidedly negative tone. It seems likely that the author has had some bad experiences in their working life. That said, there are some solid ideas mixed in that are worth at least a skim.

Recommended by someone I was interviewing for a forecasting position.

[Steffen Jensen · Rating 3 out of 5]

A great book on Forecast Value Add (FVA) analysis. However becomes quite repetitive, and even though it's a short book, it could have been even shorter without any points was lost.

[Fausto Betances · Rating 4 out of 5]

3.5 stars
Pro: plenty of good explanations. Bad: Seems to be a pre-big data work.

Following are my takeaways from the book:

Induction: reasoning from particular facts to general conclusion.
Demand: what customer want now (not what they are able to get/buy)
Bias: whether forecast are too high or too low.
Coefficient of variation: measure of demand pattern volatility. CV=Standard Deviation/mean

Volatility degrades forecast accuracy.

Inherent volatility: variation in consumption patterns
Artificial volatility: caused by company’s practices (delivery schedules, backlog, others)
Naive model: a simple forecast model with no extensive calculations.

Aim to get appropriate model over perfect fitting.

Forecast accuracy impeding factors: inappropriate model, excessive randomness, behavior change over time.

Forecast Value Added (FVA): is a metric for evaluating the performance of each step in the forecasting process.

Lean approach to forecasting means eliminating waste using FVA.

Intel’s 6 yrs analysis determined the forecast models performed same as naive models 50% of the time. When better, they were only 10% above naive equivalents.

Alternative approaches to forecasting: Statistical, Collaborative, Proactive Collaboration, Demand Smoothing, others.

Practical steps to forecasting:
1.- Recognize Volatility vs. Accuracy relationship
2.- Identify Inherent vs. Artificial Volatility
3.- Understand what level of Accuracy to reasonably expect
4.- Use FVA
5.- Eliminate worse practices

[Erin M · Rating 4 out of 5]

I had the pleasure of presenting with Michael Gilliland in October about some of the principles that are in this book. I figured I might as well read the whole thing and learn more about his opinions. I was not disappointed. For someone who works for a software company, I didn't feel like Gilliland was trying to sell me anything - I love his approach that often the simpler solution is the best, and that not every person in the company needs to have their hands in the forecasting process. His writing style is succinct, easy to understand, and sometimes even witty. I am very glad I took the time to read this book.