Most focus “on-screen” and involve strategies for infusing customer interactions with more meaning in areas like advertising, search and shopping. What remains largely unexplored is a critical back-end process— one that happens behind computer and mobile screens—namely, forecasting. After all, can a retailer or manufacturer really expect to create a relevant experience if they can’t deliver the right number of products to the customer at the right time?
Traditional retail forecasting worked like this: marketing, merchandising and supply chain each created their own guesstimates (based in large part on past sales performance) of anticipated demand. Sometimes—though not often—they would share the projections with one another, and a single number representing expected sales was agreed upon. There was little, if any, insight into the efficacy of each function.
Numerous questions remained unanswered or overlooked: What impact did the most recent two-for-one promotion have on sales? How did store-based stock levels compare with national levels after the holidays? How much were distribution issues impacting the overall sales? Which products needed more advertising, and which could achieve the same results with less?
Needless to say, these insights are as invaluable to in-store operations as they are to online operations. But until now, most retailers lacked the tools and processes to drive them into their organizations.
Agile, intelligent forecasting
With digital’s emphasis on relevance and “market of one” precision targeting, the failure to understand every link in a product value chain is no longer tenable. Understanding what stock you need where in order to meet demand can change daily—even hourly—depending on a range of variables, such as weather, holidays, promotions or competitor incentives. Reducing forecasting errors by 50 percent could not only save millions, it can also give a huge intangible boost to the brand; being seen to be available with minimum waste in an on-demand world is a socially enviable position to be in. That’s why retailers and manufacturers are increasingly turning to the more multi-faceted approach of causal-based forecasting.
Though not new, causal-based techniques have, up until now, been underutilized when calculating demand projections. If used, they have been confined within one functional group; the impact of a “buy-one-get-one” marketing promotion is only factored into next quarter’s marketing forecasts and not considered by merchandising or supply chain in their projections. In today’s digital age, having an integrated, dynamic ability to forecast demand is increasingly critical: surges happen at lightning speed. Consider the demand upswing that can occur from discount promotions around the holidays. Being able to factor the promotion into shipping and fulfillment orders accurately will have a significant impact. And things are only getting more complex. Emerging trends like Groupon and Yelp Deals, where virtually unlimited numbers of buyers can suddenly demand a given product in a compressed period of time, will create new challenges for forecasting accuracy. Aligning marketing with merchandising and supply chain through causal-based forecasting is a powerful weapon for meeting the demand in today’s mercurial retail landscape.
So what is it exactly? While most forecasting techniques rely on past performance to gauge future demand, causal-based forecasting takes a much more granular look at both internal and external factors that impact sales. Those include:
- Price changes, up or down, on a given SKU.
- Different product features such as variations in packaging, volumes and other physical attributes.
- Efficacy of promotions like two-for-one, coupons and specials.
- Advertising campaigns including TV, radio, print and circulars.
- External factors like major holidays, weather and competitor efforts.