A password will be e-mailed to you.

It is Friday evening and Taylor Swift tweets an endorsement for your product #lovinIt. Soon after, your cloud based analytical processor identifies this along with text analytics from comments using your web crawlers and artificial intelligence. It then immediately adjusts the forecast for the next 36 hours.

In addition, in areas where inventory may become an issue micro targeting campaigns kick in, automatically offering similar products at discount to shift demand. On Monday morning you come in and waiting for you is an automated report letting you know everything that occurred and the potential estimated impact on retail replenishment for the next period, what may have been cannibalized and what was incremental volume, and how this affects your financials for the year. Technology eh? The tools we have to predict demand are truly remarkable.

Forget Twitter, I’m Still Using POS Data

This is wonderful, but I live in a different world. In my world we have POS data, not Twitter comments from Taylor Swift. Our POS data is not directly tied to our planning system yet. We would love to forecast even true orders, but with system limitations and idiosyncrasies in our key customers’ ordering behavior, we rarely have an accurate picture of original customer demand. Our data is messy, and we struggle with multiple platforms that do not communicate well. I spend way too much time still educating and selling demand planning to others in the organization and explaining why the forecast is always wrong.

As great as it would be, we’re a far cry from using AI to interpret unstructed data like Twitter to produce automated, real-time demand insights and gauge impact on inventory. No matter where you are on the spectrum of maturity though, there are still some fundamental principles we all must follow. This is important – wait this is a VERY important point to understand.

Transitioning From The Basics To Advanced Forecasting & Planning

There is a lot of talk (admitting I am one of the biggest culprits) about the future of demand planning and at the same time we seem as a profession to still be stuck and not progressing forward. There are a multitude of reasons for this, but I believe one of the main ones is that we still struggle with embracing, understanding, or anchoring to a few basic laws of demand planning. Business forecasting, like everything, has its own principles that you must know if you want to use it to add value to your business. As you start from the basic world most of us live in to the advanced analytics world, I encourage you to use these principles as your guiding star and reference.

5 Unshakable Laws Of Demand Planning

Here, I would like to present the five basic and most important principles of business forecasting. These principles must be forefront in the mind of every planner and communicated often to every executive.

  • Demand planning includes uncertainty. Remember that you want to forecast the future, which is something unknown. So, you cannot expect that you will predict the future reality with 100% accuracy. Because it is expected that your forecasting will be wrong, the real question is, “by how much?” Forecasts can be wrong and demand plans can be accurate if you include probability and/or an estimate of error.

 

  • Demand Planning is less precise and less accurate as you get more detailed or granular. Product families or product groups will have less uncertainty and more accuracy than at an item-location level. For the dimension of time, forecasts will deteriorate and have more error as you go from months to weeks to days. It is more important though to plan demand at the right levels that meet the needs of the use of the forecast with the least amount of uncertainty. The way you forecast and plan demand is unique to what question you would like the forecast to answer.

 

  • Demand Planning is more precise and accurate the closer you are to the demand. Tomorrow is more predictable than three months from today or one year from now. It also stands to reason that point of sale or actual customer demand are closer to real demand than retailer or distributor orders are. Retail or distributor orders are in turn closer to real demand than shipments from a factory. The objective is to get closer to consumer sentiment and true demand. Recognize that the closer you get to the true demand signal, the better the forecast will be.

 

  • Demand Planning improves the more you know or can see. Demand planning relies on historical data and external environmental factors. Historical data is an important starting point for many forecasts. At the same time, you can’t expect to have the same results from forecasting if you do not have enough history or there is some change in environmental conditions. We want as big and complete a picture as possible – this will be better than a narrow picture from a small sample. Lots of interlocking weak information may be vastly more trustworthy than a point or two of strong information.

 

  • Demand planning is a process. With that, no matter if you need to estimate the result of 2+2 or predict the future impact of a Taylor Swift Tweet in real time, it uses the same basic steps and principles. It will require collection, transformation, modeling, and analysis of data with the goal of gaining consensus of a future forecast. Most importantly though, demand planning is a collaborative process, not a test of statistical algorithms. In many ways we manage assumptions and expectations more than we manage numbers. The analytics provide a solid foundation to work with, but the real value comes from answering questions or enabling others to make decisions.