This week I received an urgent text message from a good friend and former colleague in the consulting business:
Mike, I am down 24 pages in your book and [besmirching my trousers]. Is there no hope for consultants like me?
My response was to not let the book r… uin a good pair of pants! As long as you aren't selling snake-oil like this guy on the right, there is plenty a forecasting consultant can do to assist a needy client.
The main thing about forecasting is that we probably can't do it as well as the customer wants, so it is important to manage their expectations, determine the “forecastability” of their demand patterns (e.g., by seeing what accuracy a moving average would have achieved — and then trying to at least get a little improvement over that), and seeking alternative (non-forecasting) approaches to address their business problems.
For example, if they have lots of new products, lots of pricing and promotional activity, and highly volatile demand, then there may be no way to forecast this well, so they shouldn't build their plans around an assumption of highly accurate forecasts. But there is still plenty they can do to improve operations, minimize risks, and make more money. For example, by making the supply chain more responsive to handle demand volatility, there is less reliance on highly accurate forecasts. Or by doing things to encourage smoother (and more forecastable) demand patterns, they will get better forecasts for free, and have smoother running (and less costly) operations.
The main thing is recognition that forecasting is difficult and in many situations impossible to do as well as we'd like. It is better to approach the business problem with this recognition, rather than go down the non-productive path of expecting to achieve forecast accuracy that is impossible to achieve.
(Illustration by Jessica Crews)