From Push to Pull: Data Delivers Insights Into Consumer Demand

This article originally appeared in the November issue of CGT (Consumer Goods Technology), an Apparel sister publication.

Hanesbrands Inc. is getting a much deeper sense of consumer demand. After selecting a demand-sensing solution, this everyday apparel maker is helping to transform itself from a push-based to a pull-based consumer brand.

With the velocity demand signal repository (DSR) from Vendor Managed Technologies Inc., or VMT, Hanesbrands intends to increase sell-through performance and inventory visibility from the time of shipment through store-level scanned sales at the register. It aims to "facilitate a consolidated view of our customers across all departments," according to Pete Bobalik, vice president of technology at Hanesbrands. Bobalik also reveals that "Speed is critical when dealing with ever-changing consumer demand."

Hanesbrands currently receives POS data by SKU, by store, by week (or by day, in some instances) from retailers that make up about 70 percent of its U.S. retail business. Velocity is the DSR solution that Hanesbrands now utilizes to manage and harmonize its POS and inventory data. In order to analyze and use that data, Hanesbrands leverages Microsoft Analysis services cubes for ad-hoc reporting, Tableau for advanced visualizations, and other statistical and GIS (geographical information system) tools, according to Santiago Restrepo, director, business intelligence & analytics, Hanesbrands.

Good Proximity
Getting POS data by store facilitates a proximity analysis. "By performing a proximity analysis, we are able to calculate the distance between retailers' stores and their competitors, and understand if the performance is different," Restrepo explains. "For example, we realized that the stores of retailer X that were closer to retailer Y were performing worse than the stores that are distant. This enabled us to pay special attention to these stores and improve their performance by changing some of words of wisdom.

"Having 'Performance' and 'Potential' for each store, we can understand how a store is performing and take specific actions. For example, the action we take on stores with high potential, but low performance, is likely investing resources, while we won't invest resources on stores with low potential and low performance."

The marketing drivers such as assortment and promotions." Another useful application of proximity analysis for Hanesbrands is to understand the performance around a major mall. The challenge is that not all stores are inside the mall; for example, mass retailers are usually close to a major mall and therefore, are part of the same mall "ecosystem."

"By doing a ecosystem analysis, we were able to determine the stores that were close to the major malls and analyze the whole mall ecosystem," says Restrepo. "We were able to understand how performance was different based on the mall's composition and competition, and take the appropriate actions."

Consumer segmentation
Several processes at Hanesbrands have benefited from having accurate POS data by store, leveraging a "consumer segmentation schema," a system that classifies all U.S. households into consumer segments, based on demographic and lifestyle characteristics, and implementing advanced analytics methodologies.

Here are a few examples:

Store assortment optimization: With a consumer segmentation schema, Hanesbrands was able to create a consumer profile for a specific product or behavior, and determine the best stores for that specific product/behavior. "For example, if we are launching specific apparel for sport A, we can get a consumer profile or index of each segment on sport A," says Restrepo. "Using the consumer profile, we can analyze all the customers around every store, and determine the best stores to place our sport A products, or do promotions. Implementing a methodology like this usually increases sales per store about 300 to 400 percent."

Test and control: A traditional consumer goods company's retail tests often look at whether sales increased or not, and only at the stores where the test was performed. "Unfortunately, external factors, like seasonality, competitors' actions, etc., usually clutter the analysis and make it difficult to evaluate the effectiveness of the tests," Restrepo explains. "The methodology we developed uses a clustering algorithm to define 'similar/control' stores,
and compares their performance to the 'test stores.'"

Forecasting: Rather than basing forecasting on orders or shipments, which have a lot of fluctuations due to the retailer's inventory adjustments, Hanesbrands is changing its forecasting processes to be more based on POS, resulting in "being able to increase the forecasting accuracy and providing better service to our retailers, while decreasing our inventory levels," says Restrepo.

So what's next for Hanesbrands' demand-driven future? "We've been successful in getting the data and building the systems and processes that enable our retail and marketing analytics practices," maintains Restrepo. The next phase, and the most difficult one, according to him, is "change management."

That's when "our sales and marketing organization will embrace and utilize the new analytical tools and systems, making them part of their day-to-day job," he says. Another upcoming challenge will be to get retail buyers to understand and implement the new analytics methodologies. "Right now our focus is on embedment, training and usage increase. We've made a lot of progress, but have a long way ahead of us," Restrepo concludes.

This ad will auto-close in 10 seconds