Issues, current state and future direction of business forecasting (book excerpt)

0

Last week Len Tashman, Udo Sglavo and I announced release of our new collection: Business Forecasting: The Emerging Role of Artificial Intelligence and Machine Learning. Today, we share an excerpt from the book, one of the sixteen opinion/editorial "Afterwords" contributed by influential leaders in academics and industry, this one by Simon Clarke.

Simon Clarke photo

Simon Clarke

Simon is currently a Principal at Argon & Co, after a long career at Coca-Cola, where he served as Group Director of Forecasting. I met Simon in 2015 where he was the discussant for a paper I presented at the International Symposium on Forecasting. I was immediately impressed by his critical, yet fair and sober assessment of my paper -- this was a guy I could learn something from. I've been a huge fan of Simon's ever since, culminating in a joint presentation we did at the Institute of Business Forecasting in 2017, on the "Role of the Sales Force in Forecasting."

Today Simon and I are both Associate Editors (along with Stephan Kolassa) of Foresight: The International Journal of Applied Forecasting. Here is Simon's Afterword:

Business Forecasting: Issues, Current State and Future Direction

In Sir Arthur Conan Doyle’s, “The Adventure of the Reigate Squire”, Dr Watson takes Sherlock Holmes to a friend’s estate to rest after a previous case. Trouble, as is often the case with Sherlock Holmes, is not far away. A murder occurs at a nearby house requiring his attention. The case is solved but not before the following observation:

“It is of the highest importance in the art of detection to be able to recognize, out of a number of facts, which are incidental and which are vital. Otherwise, your energy and attention must be dissipated instead of being concentrated.”

In some respects, the practice of business forecasting has suffered from a similar lack of focus on the vital. Foundational processes are often underdeveloped or neglected. Instead there is often a focus on seeking the “smoother pebble” or “prettier shell”, all at the expense of maximizing the benefits of a well-run, organized and calibrated process.

In many organizations there remains a lack of consideration of how to most effectively leverage forecasting in the business operating model. Very often there is a clear lack of distinction between the forecast, plan and budget. As a consequence, the forecast can become easily politicized and struggles over who owns the forecast, organization structure and decision-rights come to the fore. These battles are a distraction and obscure the more important work of aligning the forecast process to support the overall business vision and mission.

At the heart of many poorly performing business forecasts are fundamental issues of solution design. Too little focus is applied on the mundane, but critical process of hierarchy design, selection of appropriate units of measure, calendars and time buckets. Failure to get these correct disables effective cross-functional collaboration and alignment and, at-worst, can drive partisan efforts to discredit or promote an alternative narrative.

Allied to issues with design can be problems with the process itself. There is often a lack of consideration of what processes the output is supporting and what detail is really required for each. For instance, a supply chain function is typically interested in the level of detail that is consumed in supply chain planning process. This is likely to be at a location and SKU level. The finance function is interested in a more aggregate level of detail. They are less likely to be interested in the lowest level detail and more interested in the level at which products are priced or costed. In addition, the “right” forecast horizon should be considered. There is no point in a 5-year ahead forecast if the decisions to be made are focused on a 12-month ahead horizon. Only once consideration of this has been made should there be focus on the forecasting method (Judgmental, Statistical or Combination) and the methods, aggregation and inputs most appropriate. Too often these decisions are made prior to understanding the goal.

Sadly, Forecast Support Systems are often seen as a “silver bullet” which has resulted in inflated expectations and subsequent disappointment in many organizations. While the quality of systems does matter, even the most capable systems will be unable to transform a broken process or design. Systems that have been successfully implemented are often found in those organizations where the selection process has been business led (not IT), often as part of a broader transformation program. Selection criteria has included technical considerations, but also as importantly usability and forecast value-add. This has often been accomplished through comprehensive proof-of-concept pilots that pitch competing software against each other on a data set relevant to the organization. With those criteria, the software that is able to support both in-sample and out-of-sample forecast evaluation, effective integration of judgmental methods, effective aggregation/disaggregation routines, inclusion of exogenous variables and point and range forecasts often rise to the top.

Despite good solution, process and systems design, sustaining change can be elusive for many. In many instances, management have unrealistic expectations, often with little consideration of the degree of difficulty of the task. Goals are largely arbitrary and disconnected from the supporting process metrics. There is also only a focus on end results, versus metrics like forecast value-add, that focus on the health of the process. In addition, the role of the forecaster has become consumed with the administration of the S&OP process, more time is being spent on assembling the data for the demand review, building the presentation and consolidating inputs and changes than is spent on essential forecasting tasks. It is very rare that the forecaster has the time to be able to review historical data, model selections and their results and experiment with alternative approaches.

A focus on fundamentals must remain high on the priorities of those wishing to advance their performance. There must be a fact-based, detailed and objective assessment of processes, people and metrics to identify gaps and assist in the development of a set of priorities. The output must provide an understanding of how mature the end-to-end process is, where improvements are required, how to structure the improvements and what benefits should be expected and tracked. Only once these improvements are made and a solid foundation created can some of the emerging and more advanced capabilities be fully leveraged.

Some the more advanced capabilities that should be expected to become more widely adopted in the future will include demand sensing, shifting and shaping. Demand sensing serves real opportunity in the near-term to anticipate and react, in close to real-time, to changes in patterns of demand. This will place increasing importance on the ability to store and process large quantities of consumption, social media and IOT data. To fully capitalize on these data sources there will be an increasing emphasis on cloud computing, artificial intelligence and machine learning.

Demand shifting and shaping is reliant on the ability to be able to manage the trade-off’s between customers and model the effects of typically price, promotion and display. This will require excellence in not just extrapolative, but also explanatory methods.  It will also require the forecaster to be able to make persuasive, factual and perhaps visual business cases of the different scenarios under consideration. With a clear distinction between short-range and mid- to long-range forecasting there will be an increasing focus on how to optimally organize data temporally and align the construction of the forecast with the intended usage.

To avoid becoming distracted with the “smoother pebble” or the “prettier shell”, and instead focus on the most impactful actions, the following should be considered:

  • Complete a full audit of the existing capabilities – process, people and metrics.
  • Prioritize the improvements based on the degree of difficulty of making the change and benefits.
  • Focus on people and process first and systems second.
  • Look to the future only once the present is secure.
Share

About Author

Mike Gilliland

Product Marketing Manager

Michael Gilliland is a longtime business forecasting practitioner and formerly a Product Marketing Manager for SAS Forecasting. He is on the Board of Directors of the International Institute of Forecasters, and is Associate Editor of their practitioner journal Foresight: The International Journal of Applied Forecasting. Mike is author of The Business Forecasting Deal (Wiley, 2010) and former editor of the free e-book Forecasting with SAS: Special Collection (SAS Press, 2020). He is principal editor of Business Forecasting: Practical Problems and Solutions (Wiley, 2015) and Business Forecasting: The Emerging Role of Artificial Intelligence and Machine Learning (Wiley, 2021). In 2017 Mike received the Institute of Business Forecasting's Lifetime Achievement Award. In 2021 his paper "FVA: A Reality Check on Forecasting Practices" was inducted into the Foresight Hall of Fame. Mike initiated The Business Forecasting Deal blog in 2009 to help expose the seamy underbelly of forecasting practice, and to provide practical solutions to its most vexing problems.

Leave A Reply

Back to Top