The Pecking Order of Demand


Of late we noticed something strange in some of the big companies that we are working with and we guess they found something strange about our observations. The demand planner in all of these companies was working directly at the SKU level and nothing above it. This means that if the company has 10,000 SKUs, the planning team directly forecasts for each of them. Of course, the good old ABC analysis brings some smiles to their faces. So when we see their forecast file, we do not see anything beyond the SKU number for thousands of rows in excel.

Something made us question our own thinking because, in all such meetings, one person found the fact unusual while the other found the observation unusual. Words ‘hierarchy’ did not seem friendly to them. This made us question our own thinking! – is it really important to make clusters in the form of hierarchy to view the data for the planner? Obviously these guys did not think so and have been doing fine internally all this while. Who is there customer – production and/or procurement, and this is all they need to get on with their job, the demand by SKU! So what is the big deal about the hierarchy?

Honestly, this did fox our thoughts for a few moments, questioning ourselves. However, as we got into discussing some of the major issues they face that leads to fire fighting in bridging the demand-supply gap, our doubts were over and faith reestablished. You really cannot (or maybe should not) aggregate the demand at the lowest level to start working on the forecast. That obviously is the end product of any system or process.

Forecasting is all about picking up the leading indicators, slicing and dicing the data, questioning the results, reiterating this a couple of times before you finally accept the number or override it. This is where the art form takes over the baton from the statistics. Is all this really possible if you have just one dimension of the data with you? Certainly not! The absence of this speaks a lot about the forecasting process and also the management view of the data as well.

When you directly view the data aggregated at the SKU level, you are making a big mistake of averaging the demand across different levels, such as product lines or category or regions etc which in turn will make your future predictions skewed too. You are most likely to end up catching up on the demand as it builds up or tones down, till it shifts the level of your average up or down. To us, you really don’t need a planning team to read the writing on the wall. The value is in being the whistleblower, in being the scout that senses the trend before anyone else does. This can only happen by adding multiple dimensions to the data.

Perhaps it may help to view the sales team as your customer as well if that helps to drive the point home better. Imagine the level of customer service you can help them achieve (and in turn the company) if the successful discovery of a demand trend in Region A is given as a heads up to Region B before it actually catches on there. That this will directly impact the bottom line as well, is perhaps another writing on the wall and hence not worth mentioning.

The source:

Too much manual work in Excel?

Streamline your planning today:

  • Forecast, plan, and place orders twice as fast.
  • 90-98% reduction in stock-outs.
  • 15-50% reduction in excess inventory.
  • 35% higher inventory turnover.
  • 10-40X ROI in the first year. 100% ROI in the first month.
  • GMDH Streamline already manages over $3 billion in inventory for retailers, wholesalers, distributors, manufacturers, and ecommerce worldwide.