Big Data Solutions for Retail – They’re Flying Off the Shelves

Today’s retail business is a real-time, information-driven enterprise. Every customer interaction and movement of a product through a distribution network is measured and used to refine pricing strategies, update inventory decisions and tailor customer incentives on websites, email and mobile devices.

The days of relying solely on POS data to determine pricing and manage inventory levels are long gone. The same goes for relying solely on newspaper ads and mailed coupons to attract customers. Maintaining long-term, profitable customer relationships requires a constant two-way flow of information between retailers’ storefronts (web or physical) and their suppliers and distribution networks. Competition is fierce and plentiful. If retailer X doesn’t have a desired item in-stock and at a competitive price, then the website for retailer Y, Z and many others are just a mouse-click away, or directions to their nearest store are easily found through a phone’s mapping app.

The best retailers have been using new mobility capabilities, Big Data technologies and advances like in-memory computing to revamp their business processes. They’re accessing information sources that didn’t exist before or were too costly or complex to use to become more nimble, cost competitive and able to delight customers with outstanding selection and service.

Spotting “Window Shoppers”
A key challenge in retailing has always been detecting and measuring lost sales. The cash register is the final system of record for all successful transactions, but what about the missed opportunities? Who was in the store or website, and what did they look at and not buy? If a retailer can understand the actions that didn’t result in a sale, they’re more likely to know if they need to adjust product selection, pricing or some aspect of displaying and promoting their offerings.

This type of information has been very difficult to track, but Big Data technologies such as Hadoop and in-memory computing are ideally suited to collecting and analyzing unstructured data types like the web logs that show the movements of every customer though an internet storefront. Web traffic data can then be combined with existing BI applications and sales data to provide new insights. For example, retailers can compare the volume of website traffic for a given product versus number of sales of that product. You’d expect a correlation between web traffic and sales – consumers find the product they want, then they buy it. If instead you find a lot of web traffic but few sales, something is amiss. It’s a signal to the retailer to keep the product (whereas in the past they may have discarded it due to low sales) and confirm the product is competitively priced and has a compelling and informative presentation, array of colors and sizes, and all other aspects that are required to incent the customer to make that final, and most important, step: the purchase.

What about physical stores? Is it possible to use the same techniques to better understand shopper behavior? The answer is becoming “yes.” Some of the leading-edge retailers are now using these new technologies to analyze video from their in-store camera systems and create mappings of customer foot traffic throughout the stores. This Big Data stream is then combined with sales data to create new applications that help optimize store layout planning product placement and uncover situations where consumer traffic (interest) doesn’t match expected sales and thus signals an issue needs to be investigated.

The following is a demonstration of an internet retail scenario where traditional sales data is combined with web store navigation to gain a better understanding of consumer behavior.

Staying Ahead of Traffic
Retailers are also using Big Data to better utilize their distribution networks and delight customers with improved on-time deliveries. In the past, dispatchers with clipboards and two-way radios would monitor daily customer deliveries and come up with workarounds to deal with traffic congestion, weather, construction and last minute rush orders. The system’s success was highly dependent on the intuition and expertise of the dispatcher. Today there are ways to make dispatchers job’s easier and augment their decision-making with information from real-time data sources.

Radio transmitters on every truck along with bar codes or RFIDs on each package are combined with real-time mapping and traffic information to allow dispatchers to better monitor and visualize the progress of every delivery. Literally in the first minutes of the day, after just one or two deliveries, a predictive analytics application is already providing the dispatcher with revised estimated delivery times for the remaining orders based on past delivery data and current real-time traffic data on every truck’s route. If a dispatcher predicts a truck will miss a delivery, they can take immediate corrective action, such as re-routing a delivery or rescheduling with a customer. This new ability to provide organizations with easy-to-use applications that map incoming customer orders, real-time traffic and current truck location information has allowed leading organizations to do a much better job of meeting customer expectations and ensuring high operational efficiency in their distribution network.

An example of a world-class organization that has implemented a number of these techniques is the online grocer FreshDirect. Check out this video to hear more about how they are using better information and real-time insights to deliver exceptional customer satisfaction, and of course groceries.

The Road Ahead for Retail
And wait, there’s more to come. The holy grail of retail has been to anticipate what consumers need even before they realize they need it. There’s no better way to beat the competition than to make an attractive offer and get a customer’s business before they even realize they need your product, or consider evaluating alternatives. Take printer cartridges, for example. There’s nothing worse than having to print a boarding pass with the taxi waiting outside and realize you’re out of printer ink. Today, retailers of office supplies are able to track purchases of customers’ in-store credit cards and rewards cards and, based on purchase history, anticipate when a consumer might need to reorder a product. Today they can send out an email offer for printer cartridges as well as an accompanying promotion for paper, with a guaranteed delivery time of 24 hours. Similar techniques are being used by travel companies (Time for your annual vacation?) and auto dealers (Looks like your car is due for service.) as well as other consumer-facing organizations.

Moving forward, imagine a world where retailers can use these new data sources to expand their consumer intelligence base to include analysis of customers’ social environments and web-patterns to become even more relevant and anticipatory of needs and interests. For example, an office supply store could not only provide you with an offer on printer toner but also show you who in your social network “Liked” that product and what else they bought. Or, they might indicate that friends in your social network have gone to a vacation spot you are exploring, and that they had a great time – or not. Clearly there is a need to manage the privacy aspects of this, but if retailers can provide an appropriate “opt in” and ways for consumers to manage the flow of information that is shared and who sees, it could be a great opportunity for retailers to enhance customer service and loyalty by providing consumers with better advice and less impersonal spam and junk mail.

I will end with a reminder that SAPPHIRE NOW + ASUG ANNUAL CONFERENCE is next week (May 14-16) in Orlando, where SAP, our partners and customers will be discussing (and announcing) more Big Data concepts and how retailers and all other industries are leveraging these exciting capabilities to redefine their businesses. I hope to see you there!