A Whirlwind of Unpredictability
For over a year now, manufacturers selling mainstream products to retailers in virtually all tiers of distribution from mass to specialty have ridden a whirlwind of unpredictability. How do they make inventory decisions in such a disruptive environment? They can plan based on history, but how relevant is it now? I am commonly asked what methods of forecasting I recommend, and which are the most accurate. My answer is a methodology based on POS and a sales curve. While our methodology hasn't changed, how we use the parameters has. In addition to POS and a sales curve, we integrate a weeks of supply model, lead time and adjustment factor. The adjustment factor takes into account any "outside" information that the pure data does not reveal- such as changes in merchandising strategy or increase/decrease in store count. And in this environment, "outside" information might include non-merchandising facets. Like the fact that there are simply less consumers in stores. But how do you quantify that? One way is to follow key economic indicators. The Consumer Price Index, new home starts, inflation, unemployment, the Dow Jones Industrial Average and industry specific indicators like the Computer and Electronics Index may help us quantify it. If we could prove that a 1 cent change in the price of gas related to a 2% decrease in sales we could adjust our sales curves and adjustment factor based on the expected price of gas. Can we use this information to improve our predictability model or is it too generalized to affect a specific product at a specific retailer? A recent client project proved to be very interesting and I thought I would share it with you.
First we graphed sales by week for 2009. Then we graphed a variety of economic indicators to see which ones followed the same pattern. The following charts show the relationships between 3 economic indicators and sales history for our client's item (client name, item and category have been masked here for confidentiality reasons).
While the peaks and lulls of sales of the item were more dramatic than the index, this example clearly illustrates that both follow the same trend. This would indicate this item is indicative of the industry as a whole, and future estimates on the index could be used to help adjust our curve up or down.
Again, in this example sales for the widget follow the same general pattern as the Consumer Price Index. A promotion was run during the Holiday selling season which does skew it a bit, the peaks and lulls are closely related.
Finally, we compare the Dow Jones Industrial average against our sales. In this example it's hard to make a case that the Dow would be a useful source for predicting future sales.
In addition to looking at your sales, you probably rely on gut- feel one way or another about the future state of the business. However, adding a more scientific approach can provide more confidence in your decision making. With a couple of quick Internet searches you can get the forecast for most of these indicators for free. If you can determine that your item consistently follows a similar trend as a particular economic indicator, you might consider adding it as a variable when accessing risk when planning your inventory needs.
While this may seem a bit far-fetched, many large retailers use this type of data to project their total company sales (and offer stock performance guidance), which in turn gets pushed down to division, class, department and ultimately the open to buy for your items. It doesn't take into consideration the promotional activity planned or presentation standards, but during a time when you are trying to gauge what the new baseline is, it may be helpful. So unless your products or brand carry the same cache as Apple, retailers will expect you to improve your forecasting skills in magical and revolutionary ways.
I would be happy to share our perspective with you, or help you strategize. If interested, please call me at 212.938.1991 x101 or email me at jlewis@ers-c.com
