By Ron Menich, Ph.D., EVP and chief scientist, Predictix
Wherever you turn today, Big Data is impacting every facet of society and business — and retail is no exception. For example, consider the impact of weather patterns on retail sales.
Cooler-than-normal spring temperatures might result in lower sales of categories such as garden care and outdoor living. However, if retailers don't take these abnormal weather patterns into consideration when planning for next year, they may seriously underforecast and end up with out-of-stocks — all because they failed to take advantage of available weather data and adjust their future forecasts accordingly.
Big Data alone is not enough: The three keys to consumer insight
In all its many forms, Big Data has the potential to provide retailers with a clearer, more accurate picture of consumer demand for forecasting systems. However, Big Data alone is not enough.
There are two primary barriers that can prevent retailers from fully leveraging the power of Big Data. First, many retailers are currently running outdated forecasting systems that lack the ability to quickly and effectively analyze Big Data; after all, they were designed before the concept of Big Data even existed. Those older systems get around the limitations by taking shortcuts and making analytical compromises. Moreover, retailers can't afford the IT infrastructure that's needed to process staggeringly large amounts of disparate data.
Getting the best demand forecasts using Big Data requires a combination of Big Data, predictive analytics and the cloud, all working together to bring consumer behavior and demand into sharp focus. The on-demand resources of the cloud provide retailers with unprecedented scale and ability to manage, mine and process data volumes that were previously unthinkable.
And the unlimited resources of the cloud are ideal for highly sophisticated, configurable and up-to-date forecasting science that utilizes all kinds of structured and unstructured data. The term “the cloud” is used so much these days and “cloud computing” has a number of differentiating qualities, so I want to be specific by what I mean by the term.
For purposes of this column, the primary use of cloud is “elastic computing,” meaning the rapid deployment and use of computer resources which vary dynamically to meet a variable workload. By utilizing this elastic computing capability, a retailer can process extremely large amounts of data when it needs to do so, and then relinquish those computing resources (and their associated costs) until the next time they are needed.
Big Data in action: Improving forecast accuracy on 70% of promoted items
As an example, one Top 10 U.S. retailer uses a combination of Big Data analytics, advanced forecasting science and the cloud to accurately forecast store-level promotional demand. By harnessing the power and elasticity of the cloud — calling up as many as 400 servers at a time — to perform forecasting and analytics and process Big Data, the retailer’s buyers and merchandisers benefit from improved real-time decision making. They can now generate forecasts on-demand for any of more than 50,000 items across more than 8,000 stores in about a minute. These types of processing tasks formerly took many hours using traditional in-house computing resources.
The result? This retailer now operates with higher forecast accuracy on 70% of promoted items and a 25% increase in store-level inventory productivity — a dramatic improvement that has transformed promotional forecast accuracy.
The Big Data imperative
The volume, velocity and variety of data sources in retail are increasing dramatically. And the good news for retailers is that, for the first time, it's possible to truly leverage insights from all kinds of Big Data, in combination with advanced predictive analytics and the unlimited potential of the cloud. Retailers can also do it quickly and cost-effectively, making it possible to drive truly big decisions with major impacts on profitability. Big Data is here to stay.