Turning the Kaleidoscope of Data to Discover Value

Access to point-of-sale data has transformed the retail businesses by suddenly throwing open new insights from data that were hitherto not captured. This knowledge has empowered the retailers with an ability to understand their business better and use these insights for accurate decision-making.

Various statistical techniques can be used in the right combination to analyze data and arrive at trends and patterns that lead to increased sales and thereby directly impact the bottom-line and result in profitability. Using these techniques, retailers can categorize customers by the products and services they choose, identify patterns to plan cross-selling campaigns, analyze and target customers based on product-centric purchase histories and patterns, plan multi-product promotions based on customer response, arrive at customer probability to buy additional products, measure shifts in customer behavior and locations, compare segments the possibilities are endless. 

Realize the connection

Retailers are aware of the fact that shoppers who buy one product frequently, (for example, hamburger patties) are more likely to buy a couple of other related products (for example, hamburger buns and fries). This probability is due to the amount of affinity that exists between the products.

While some product affinities, such as the one above, are obvious to observers and can be arrived at by using common sense, others can be less apparent without the appropriate data mining techniques; for example, Wal-Mart customers who purchase Barbie dolls have a 60% likelihood of also purchasing one of three types of candy bars (Forbes, September 8, 1997). This paired purchasing pattern can be attributed to the obscure correlation between the two products.

Every shopper’s basket has a story to tell

The items a shopper purchases during one shopping trip makes up the market basket. The various items in a market basket are correlated to each other with varying frequencies, presenting a picture of what may have driven the shopping trip: running short of a couple of essential ingredients for cooking, assembling all necessary items for an exotic meal or a birthday party, preparing to have guests over, stocking up groceries for the month

The selection also reveals the shopper’s profile to a certain accuracy and provides a glimpse of socio-economic attributes of the shopper: is the shopper a family person? Is there a child in the shopper’s family? Are there elders in the family? What age-bracket does the shopper fall into? Is the shopper an impulsive buyer? What economic section does the shopper belong to? The overall level or value of the selection is another factor that can be of significance for retailers. Such information when viewed on the time parameter reveals stronger correlations: among products, between shopper profiles and their product preferences, between the kind of shopping trips and the product preferences, between the time of the year and product preferences of a particular shopper profile, to name a few.

How business analytics can help

Applying these seemingly simple statistical principles of great practical use to the humungous amount of data in any large format retail is a Herculean task. Sifting through this data and reaching any valuable conclusion is quite like looking for a needle in a haystack.

Business intelligence (BI) systems are often looked at as a solution to this challenge. However, the piece that separates a generic BI system from the one that facilitates retail decision making is advanced analytics. BI systems designed specifically for the retail industry and powered by advanced analytics can help drive the functional theories of mathematics in retail decision making. The analytics angle brings in a scientific base for choices that were made randomly or based on observation and experience earlier, guaranteeing a considerably higher success rate. The prescriptive and guided nature of retail-centric BI systems, along with their out-of-the-box functionalities, makes it possible for unskilled personnel in the retail chain as well to use these statistical techniques in their area of operations to arrive at benefit-driven decisions. An attempt to assemble a BI tool-kit for such analytical requirements can prove hazardous leading to uncertainties and perils involved in the never-ending process of adding blocks to map newer business demands and analytical needs.

Retail BI applies the principles of product affinity and market basket analysis to point-of-sale data to reveal concealed relationships between products and discover customer behaviour patterns. Such analysis provides insights into the types of products customers usually buy together, the time of year when the sales for a combination of products increase, destination items that attract customers to the store, and reasons for a sudden boost in the sales of a specific product. The analytical abilities of retail BI can identify correlations between customer profiles and product purchases and store visits.

With the help of retail BI, product affinity and market basket analysis can be used by the marketing and merchandizing teams at various levels in a large format retail scenario. Market baskets can be profiled and classified into categories such as grocery basket, special occasion basket, weekly shopping basket based on the objective of the shopping trip as revealed by the analysis. This information can be used for planning daily promotions to drive more trips to the store by offering a discount voucher for a week or to increase the basket size during each trip by reducing prices of certain products or offering special discounts/gifts. Similarly, product affinity data can lead the store manager to identify the products that most shoppers look for in the store and place them in an easily accessible area or identify the products that are bought together most often and display them in close vicinity or arrange the products identified as impulse purchases in an attractive manner to make the impulse irresistible.

Decisions related to the selection of products on discount, special offers to a particular segment of customers, gift vouchers on certain products must be based on scientific data in order to derive maximum value from such initiatives. The trends and patterns that are revealed by retail BI can be used for:

  • Improving the effectiveness of marketing, sales and merchandising strategies
  • Planning strategic initiatives such as periodic promotions, campaigns, special offers, price changes, cross-selling, product-pairing
  • Correlating store performance with overall market performance
  • Planning the store layout for more effective product placement and shelf presentation with appropriate prominence for impulse purchases, seasonal purchases, destination/anchor products

Retail BI can turn the kaleidoscope of data in infinite angles leading to endless possibilities that help analyze and use the data to arrive at trends and combinations in retail decision making.