Are there some things you wish your organization’s management knew about forecasting?  Those of us who have served time in the forecasting profession know that “serving time” is an apt description of the job.  Being a business forecaster is sort of like being in county lock-up – without the benefit of free meals, charming bunkmates, and periodic delousing. 

Forecasting is difficult – we never seem to forecast accurately enough to please management.  And forecasting is thankless – even when we come up with good models that forecast reasonably well, someone above us is likely to change the numbers to whatever they darn well please.  Those forecasters that aren’t already on mood altering substances probably should be.

What the forecaster really needs are the tools to educate management and forecast as accurately and efficiently as can reasonably be expected given the nature of your demand.

Reasons for Forecast Error

There are four main reasons why forecasts are wrong:

• Unsound or misused software
• Unskilled or inexperienced forecasters
• Politicized forecasting process
• Unforecastable demand

The best accuracy you can achieve is limited by the forecastability of your demand patterns.  So accuracy expectations have to take that into consideration.  The naïve forecasting model is the proper baseline for accuracy objectives, and industry benchmarks should never be used to set accuracy targets.

New product forecasting is an area of particular angst.  Managers realize that these forecasts are usually way off, yet they forge ahead with supply and revenue plans in full confidence.  We suggest that assessing uncertainty and risk is more useful than forecasting alone. When management has a good understanding of the likely range of new product demand outcomes, the organization can better align resources to all the possibilities.

We also support Forecast Value Added (FVA) analysis – a method now used by many major organizations to identify forecasting process waste and to achieve better forecasts.  FVA evaluates every step and participant in the forecasting process, identifying those that are not adding value by making the forecast better.  Many process activities are found to be making the forecast worse – and these activities need to be fixed or eliminated.

Intel extensively uses FVA analysis. Over the last three years, Intel has taken the basic idea of FVA and applied it to a broader range of forecasting and supply chain process issues. Intel has gone through paradigm shifts in thinking, and how to address the change management issues.

Final Thoughts

On Monday February 22, at the IBF’s Supply Chain Forecasting Conference in Phoenix, perhaps we (Emily Rodriguez, Program Manager at Intel) and Michael Gilliland at SAS) can help improve your forecasting performance. We are delivering a morning workshop entitled “A Primer for Management: Fundamentals of Business Forecasting and Conducting Forecast Value Added (FVA) Analysis.” 

The theme for our presentation is “what management must know about forecasting.”  We will be providing step-by-step instructions for gathering, analyzing, and reporting the data needed for a thorough FVA analysis, along with several brief case studies at organizations, where it has been applied.  Furthermore, Emily will provide an in-depth case study of the use of FVA analysis at Intel.

Whether FVA is a new concept to you, or you are an experienced practitioner of the approach, we look forward to having you join us in Phoenix.  Meantime, stay abreast of the latest innovations and defamations in forecasting, in The Business Forecasting Deal.  See you in February at the IBF Supply Chain Forecasting Conference.

Agree or disagree with the author(s)' perspective? What would you add? Let us know your thoughts for publication in the SCDigest newsletter Feedback section, and on the website. Upon request, comments will be posted with the respondent's name or company withheld.