Optimizing Cash Flow with SKU-Level Forecasting
By Dan Raftery, president, Raftery Resource
Network, Inc. and Aaron Raftery, project manager,
Lime Energy, Inc.
Several supply chain
specialists at the recent IE Group CPG Forecasting and Planning Summit
in Chicago offered top-level peeks into their demand forecasting methods.
Paul McMorrow, senior supply planning manager
at H. J. Heinz Company, highlighted the new imperative for this discipline:
to free up internal cash flow in response to the tightening credit market
and other forces in the current economic climate. Mr. McMorrow and other speakers outlined strategies to reduce
inventory-related drains on cash flow by increasing SKU-level forecasting
accuracy through SKU fragmentation techniques.
Alan Fang, vice president,
business process re-engineering & applications, Easton-Bell Sports,
Inc., offered a matrix framework that crosses A-B-C SKU productivity against
X-Y-Z forecast accuracy, creating a grid of nine
"buckets." The AX bucket would contain SKUs with the highest sales
volume and greatest forecast accuracy. An example of how forecasters can
use this matrix: to reduce inventory levels of the C SKUs (lowest sales volume),
forecasters can focus on products in the CX bucket (slow-movers with the
most reliable forecasts).
Jan Steuber, demand planning, supply chain, MARS, presented a
similar SKU segmentation methodology. However, instead of fragmenting SKUs
using forecast accuracy, MARS crosses SKU productivity with X-Y-Z demand
variability. Mr. Steuber showed that demand variability
has a significant negative correlation to forecast accuracy, indicating
that these two variables capture similar information. An advantage to using
demand variability is that it also provides an entry point for reducing
forecast error. By adopting marketing strategies that incorporate demand
variability, Mr. Steuber suggests businesses
may be able to indirectly improve the accuracy of their forecasts.
Discussion Questions:
What advantages do you see in using SKU segmentation in forecasting?
What are the "watch-outs?" Are CPG supply chain forecast experts
really being used for critical business strategies such as cash flow
optimization or is this a lightly tapped resource?
SKU Segmentation is a terrific idea. But let’s take a more shopper-centric view. There is more to this than forecast accuracy. The SKU grading should also take into consideration which items tended to be present in the market baskets of a retailer’s tier one customers. Another segmentation technique might be “key item”–those items which are taking a primary position in the planogram. Key items can be inferred by taking the best-selling, high-margin items from the last 13 weeks of planogram performance.
This focus on individual SKUs is the flip side of the other thread today on loyalty cards. There really are only three mechanical ingredients to an individual sale: the shopper, the product and the place in the store where the connection occurs. Segmenting shoppers and segmenting products are both vast steps forward over weekly roll-ups of sales data. Both are necessary as a part of the “Return to Personal Selling at Retail,” about which I will present a keynote at next months Shopper Insights conference.
However, both SKU segmentation and shopper segmentation are still blunt instruments. The real potential is to eventually skip the segmentation and go directly to item management and shopper management, a la Amazon (not that they don’t use segmentation.) But the point is that it IS becoming possible to treat each shopper as an individual, just as Amazon adapts their offer to the individual as soon as the shopper is identified (shows up on the site) and begins clicking, giving specific, real-time evidence of what they might be seeking.
As long as loyalty cards continue to be thought of as a clever way to PAY shoppers to buy, they will continue to feed the misguided attitude that money is the most important thing to the shopper, even though we know that in CPG/FMCG world, time is typically more important than money.
How to get to the ultimate in segmentation, where one shopper is matched to one product, one at a time? That is a process that begins in the mind of the marketer and will never end, but will accelerate with the increasing deployment of the internet on “PDAs” in the store, the “Amazonification” of the store. However, long before this tidal wave engulfs retailing, smart retailers will attack the greatest barrier to accelerating sales. It is that gray matter between the retailer’s (and brand supplier’s) own ears.
It all sounds good, but reality is that most companies don’t leverage their POS/SKU level sales data very well at all. Companies might leverage POS data on a retailer by retailer basis. But in 95% of accounts that I work with there is very little POS data integration in existance. Companies know they need it, but are not investing in the necessary architecture to cleanse and harmonize the POS data. This would allow them to integrate it with their internal data and other syndicated data and apply formulas that would best fit their business. Each company needs to use the forecasting method they deem best for their product, but the bottom line is that without reliable data to report from, the reports will not be reliable.
Good article Dan, and interesting comments Herb.
One minor point. Tough to reduce inventory unless the manufacture is willing to put the product in smaller cases, with fewer products.
I think SKU segmentation based on demand and/or forecast variability is a good start.
The challenge we see is that each function is taking a parochial approach to utilizing POS data. The situation brings to mind the parable of the six blind men and the elephant. You’ll recall one man was touching the trunk and thought it a snake, another the ear and thought it a leaf, the leg a tree, the tail a rope. Today we have sales teams, sales operations, trade promotion, demand planning, VMI/CRP team and logistics each trying to solve the problem on their own yet each is falling short in developing a compelling value proposition. Moreover, manufacturers are at risk of losing their scale economies — with shopper insights, with data management, with tools and best practices–in addition to a consistent view of demand to maintain manufacturing scale.
Industry leaders need to shape a corporate strategy for POS data. They need to ‘draw the elephant’ and align disparate initiatives against a shared vision and right to win the POS game. This is a fast-changing area, so we are not ready to cast the elephant in bronze just yet. But to paraphrase the ancient Chinese proverb: If you don’t know where you are going, all roads lead there.
I’ve always believed that as essential and important as demand forecasting was, it was what you did with the forecast that really made the difference. The devil was in the execution. No matter how ‘accurate’ a forecast was, it was still highly unlikely to truly be considered accurate very often.
This, however, is a concept that is exciting. It’s an intriguing concept, that while not reducing the importance of execution, certainly provides a far more reliable starting point. I’m eager to learn more.
An awful lot of volatility stems from promotional tactics, both competitive and otherwise. Translation: coordinate your trade promotion calendar with your competitors’ calendars. In most cases, it doesn’t mean you need to communicate directly with your competitors, since most grocery brands simply repeat the same promotions very predictably, year after year. Is it a surprise to see hot dogs and hot dog rolls promoted around July 4th? Is it a shock to see turkeys, cranberry sauce, and pumpkin on deal before Thanksgiving? Suppliers, supermarket managements, and shoppers all know the promo calendar. Just once, wouldn’t it be fun to be surprised?