Get fresh updates from Hortonworks by email

Once a month, receive latest insights, trends, analytics information and knowledge of Big Data.


Sign up for the Developers Newsletter

Once a month, receive latest insights, trends, analytics information and knowledge of Big Data.


Get Started


Ready to Get Started?

Download sandbox

How can we help you?

* I understand I can unsubscribe at any time. I also acknowledge the additional information found in Hortonworks Privacy Policy.
closeClose button
March 25, 2015
prev slideNext slide

HDP for Retailers: Common Data Challenges

Changes in technology and consumer expectations create new challenges for how retailers create a single view of their customers, predict consumer preferences and discover fine-grain detail to manage their supply chains. Of course, these have always been important goals for retailers, but it used to be easier to understand the market signals (because there were fewer of those) and also to choose the right actions to take (because there were fewer possibilities there too).

Challenges Building a Single View of the Customer – Fragmented Data

In today’s data-rich world, retail enterprises need to rethink traditional ways of knowing their consumers. It is now rare for a shopkeeper to greet customers on a first-name basis when they walk into the store. Rather, today’s retailer must interact with consumers across multiple channels at once; including online, mobile devices, phone, social media and in stores.

Traditionally, retailers analyze sales data for information on purchases in one particular channel, generally store sales from cash register receipts. Additionally, web channel managers capture and analyze web data in order to determine “hot spots” or dormant areas online that are not producing sales. Customer service and call center managers collect data on sales made over the phone and on common consumer requests or complaints. The issue with managing all of these channels is that few retailers have visibility into sales and service experiences that move across channels, and so they don’t actually understand exactly what led up to each individual sale.

Retailers want a single view of their consumers or a “customer golden record”, but data fragmentation can block that. Without this single view, retailers have a hard time accurately calculating lifetime customer value (LCV) and then they cannot use that as they create promotions or offers. Without an intimate understanding of their customers, retailers miss up-sell and cross-sell opportunities.

Challenges Predicting Consumer Preferences – Scarce Data

Technological change also disrupts the retailer’s ability to market and promote their products. Today’s consumer has more choices, yet many advertising or marketing systems were designed for a simpler world where consumers were exposed to ads through one of four television networks, maybe two radio stations and a daily newspaper. This created relatively simple targeting models and led to the situation described by John Wanamaker, the father of modern advertising, when he said: “Half the money I spend on advertising is wasted; the trouble is, I don’t know which half.”

Wanamaker was describing the dilemma of scarce data: even the best targeting models would be imperfect if they lacked sufficient data on consumers. Without sufficient data on marketing efforts tied to purchasing behavior, retailers will be unable to create confident predictive analytics for customer segmentation or value-added services like recommendation engines and personalized web experiences.

Challenges with Supply Chain Management – New Types Data

Another challenge in the retail industry has been that data storage platforms could only store structured, curated data. This “clean” data did not fully describe the complexity and variability of the roots of retail demand. It may give a fairly clear picture of what consumers bought, but it poorly conveys what people thought, felt or said about the retailer’s products, service, stores, or reputation. In the hyper-competitive retail market, these intangibles can be the missing ingredient in determining what consumers are likely to do next.

With HDP and new sources of data, our retail subscribers can better predict demand for specific products at the store level. They use that insight to manage inventory and avoid stock-outs (or to avoid overstocking unnecessary inventory). With new data discovery capabilities, sentiment analysis and machine learning algorithms, retailers can more accurately plan and manage just-in-time delivery of the merchandise mix most likely to meet consumer demand when and where that demand occurs.

Hortonworks subscribers in telecom, insurance, healthcare, manufacturing, and financial services partner with Hortonworks for the same reason: to build advanced analytic applications for a single view of their business, predictive analytics and data discovery. We bring these and our direct experience partnering with some of the world’s largest retailers to each new retail enterprise who subscribes with Hortonworks.

Retailers Use Hortonworks Data Platform to Meet Common Data Challenges

At Hortonworks, we’ve seen customers in the retail industry improve revenue, maintain margins, and enhance customer service through a number of data-driven initiatives. By implementing a Modern Data Architecture (MDA) with Apache Hadoop, retailers are assembling a single view of their consumers or operations, creating targeted offers and recommendations, and optimizing their supply chain and inventories. They are leveraging advanced analytic applications to improve visibility of data already under management.

Learn More About HDP for Retail

About the Author

Eric Thorsen is the Hortonworks General Manager of Retail and Consumer Products. Eric brings over 20 years of technology expertise focused on the retail and consumer products industries. Prior to joining Hortonworks, Eric was with SAP, focused on the business impact of technology solutions in those industries. Follow @EricThorsen or link to


Leave a Reply

Your email address will not be published. Required fields are marked *