VOLUME 5 • ISSUE 1 2012
A bi-annual journal produced by Cognizant
Innovation Reimagined REWIRE
Fact-Based Insights Maximizing the Returns of Big Data
●
Fact-Based Insights
REWIRE
Maximizing the Returns from Big Data The rise of social and mobile computing means huge volumes of precious customer and prospect insights are available for the taking. Extracting and making sense of this raw data, as well as data from traditional systems of record, requires definitive use cases in which tangible business objectives drive experimentation with new tools, analytical techniques and operating processes for pinpointing potential returns on information — and investment. By Debasish Mukherjee and Karthik Krishnamurthy
Data, data everywhere and not the least bit of insight. The phenomenon known as big data1 is exploding across every industry sector, creating both challenges and opportunities for companies large and small, industrial and consumer-facing, worldwide. One driver behind the data deluge is the push by many companies to better understand customer behavior. Progressive companies are sifting through the burgeoning array of unstructured and semi-structured data contained in social media and mobile commerce platforms, as well as the growing volumes of structured data housed in traditional systems of record, for a glimpse of the future. They seek to understand not just customer buying behavior but also their own interactions with customers to inform move-forward strategies and extend growth. Another big data driver is the proliferation of sensors in nearly every industrial business sector, from manufacturing through utilities. For example, the energy industry is moving from monthly readings of consumer electricity and water usage, to smart digital metering to help consumers plan energy consumption in real-time and power generation companies to more efficiently adjust operations during daily/seasonal peaks and valleys. And get this: It is estimated that about 90% of available data has been generated in just the last two years. In fact, IDC2 estimates that data volume is growing at 50% per year, or more than doubling every two years.
Big Data and the Data Deluge: Challenges and Opportunities That businesses are swimming in data is nothing new. Digital data has been exploding since 2000, when it accounted for roughly 25% of all information stored; by 2007 its share increased to 94%. In fact, the amount of data stored has grown from one exabyte in 1960 to almost 1 zettabyte in 2010. That figure is expected to soar 44 times to nearly 35 zettabytes by 2020, according to IDC. The proliferation of digital data is fueling the need for innovation across the discipline of enterprise information management. Semi-structured and unstructured data generated by social networks and the
http://cognizanti.cognizant.com
COGNIZANTi
3
proliferation of mobile devices is growing at a faster pace than the structured data contained within traditional systems of record (see Figures 1 and 2). The emergence of consumer, company-sponsored and media blogs, tweets, Facebook chats and m-commerce transactions is spurring the need to store, process and make sense of this data, which is rich in strategic insight. This change in the variety and complexity of data has left companies feeling overwhelmed. Moreover, the hyper-competitive global economy places a premium on making decisions faster and with better precision. Data captured from social networks and mobile devices are time sensitive, not only when it comes to capturing it but using it to inform efficient decision making. Capturing quick quality insights from this data is therefore critical to maintaining or extending competitive advantage.
Data Sources Impacting Growth Mobile
Texting
Internet of things
Social
(where all devices are IP addressable)
Machine-generated data is projected to rise from today’s 200 exabytes to 1,000 exabytes by 2015.
FIGURE 1 Source: International Data Corp., "Worldwide Big Data Technology and Services Forecast," 2011.
The Reality: Massive Data Growth Stored Digital Information (exabytes)
• Multi-structured data content is the primary driver of new data.
1970
Complex, Multi-structured
Business Transaction Data 1980
1990
Web Application Data Relational 2000
2010
• 80% of this new data is digital, which is complex to analyze in its native structure. • Digital data is growing at 62% annually vs. structured data at 22%.
FIGURE 2 Source: IDC, “As the Economy Contracts, the Digital Universe Expands,” May 2009.
Established businesses relying on decades-old, but successful, data models often struggle to incorporate reams of unstructured data into their established enterprise information management infrastructure. Moreover, they face threats from newer, more nimble and technologically-savvy companies that can leverage the vast pools of proliferating data and create services and products that do not exist today.
4
COGNIZANTi
http://cognizanti.cognizant.com
Managing Big Data Efficiently: A Problem of Perception Sadly, data is expanding with minimal governance and management. A lack of expertise and usable enterprise tools, combined with the complexity of big data implementation and usability, is undermining businesses’ ability to scale their infrastructure, processes and strategic planning to meet the challenge. Various tools and technologies have emerged to store, aggregate, analyze and visualize big data. But because these tools and techniques originate from several fields, including applied mathematics, statistics, economics and computer science, organizations seeking to derive value from big data need to open their minds and adopt a flexible, multidisciplinary approach. The complexity becomes evident when you consider even a simple three-layer model of storage, data processing and analysis that encompasses many different tools and technologies, each addressing a specific problem at hand. A mix of platforms must coexist to combine structured and semi-structured storage. Data processing is performed using distributed techniques and analysis. Visualization techniques range from SQL-like tools to newer data access and analysis tools. Moreover, extensions to traditional business intelligence and statistical tools are released by the day, which makes it tough for most organizations to keep pace. (See page 85 for a glossary of tools and techniques to consider.) Traditionally, companies have used siloed solutions from third-party vendors to manage function-specific data queries. This approach often prevents them from extending the right experience or offer to the right customer and undermines customer satisfaction and business objectives. However, big data solutions (including the consolidation of structured and unstructured data) along with specialized tools and techniques offer ways to enrich customer experience with a personalized approach that enables true customization of services.
Even before they make tool selections, they must develop proofs of concept, or use cases, in which an ecosystem of business partners and customers are aligned to help broaden the accessibility to and analysis of useful structured and unstructured data that can generate meaningful insights. While tools are an important foundational element, companies need to rewire their thinking as well as their infrastructure to effectively tap the potential of big data. Even before they make tool selections, they must develop proofs of concept, or use cases, in which an ecosystem of business partners and customers are aligned to help broaden the accessibility to and analysis of useful structured and unstructured data that can generate meaningful insights.
How Big Data Drives Business We believe winning organizations are those moving forward with big data via exploratory activities. For instance, we recently worked with a consumer goods company to create negotiating leverage in its supply chain. We built a customer sentiment analytics solution based on data pulled from social media sites such as Facebook and Twitter that provided consumer insights on the company’s products and their buying experience. On deeper analysis, we found that the most important negative sentiments were related to customers’ perceptions of a poor in-store experience, such as a lack of product availability or shipment data. These views are rarely expressed in person and could only be captured by mining social media conversations. We captured and thoroughly analyzed these expressions by using our homegrown integrated sentiment analysis and reporting tool, iSMART. This helped our client determine key pain points, understand how to effectively address these challenges and leverage the entire supply chain to improve the consumer’s in-store experience.
http://cognizanti.cognizant.com
COGNIZANTi
5
This approach could easily be applied to other business sectors, particularly those where data generation is significantly higher and growing exponentially. These include media and communications, process manufacturing, government and healthcare. Financial services sectors that include banking, insurance, securities and investment services also have more data stored per company than any other, on average. Here’s how some of these sectors can exploit big data: ● ●
Financial services: Early-warning fraud detection systems, identity threat management. Life sciences: Physician interaction optimization, proactive manufacturing surveillance, next-generation genetic sequencing.
●
High-tech: Smart data management, sentiment analysis.
●
Insurance: Claims fraud monitoring.
●
Healthcare: Cross-channel analytics, campaign optimization and fraud detection.
Big Questions and Challenges With all the attention on big data, the big question is whether the trend is overhyped and if, or when, the investment required will translate into business value. Some companies are already aggressively sifting and interpreting big data and developing new business models from it. As with every new technology trend, it will take time to identify relevant and appropriate use cases. However, the process of experimenting, prototyping, iterating and validating results is crucial to deploying solutions that prove the value of big data analysis. (See sidebar for indicative scenarios.)
Quick Take The Case for Big Data Analytics The following are use cases to better illustrate how organizations can apply big data analytics to connect with customers and create competitive advantage:
6
COGNIZANTi
•
Use case 1: Dynamic pricing for mobile phones. Operators facing bandwidth congestion could consider dynamically adjusting price as a function of the bandwidth used by each consumer. For example, a consumer watching a video stream at peak hours could be alerted and automatically charged more if he or she continues to watch (after being alerted). If the user chooses to continue, the device will generate more value, and the population using the mobile device will shift toward heavy users. As a consequence, a pricing plan for “regular” users may include a clause for service degradation during peak hours.
•
Use case 2: Utility companies analyzing consumption patterns in real time. Applying the same principle as above, per-household pricing can be adjusted based on energy consumption detected by smart meters, or sensors. Valueadded service could be offered to help individual households rationalize energy usage during peak hours and offer more attractive prices to those that conserve when it is advantageous to the utility.
•
Use case 3: Consumer goods companies applying social media analytics. Most consumer goods makers do not have direct access to consumers, as they work through distribution channel partners (middlemen, retailers, etc.). Through social media analytics, they could, nevertheless, understand what their end-customers think about their brand and can better adjust pricing strategies, quality of products, etc., as the earlier consumer goods case study suggests.
http://cognizanti.cognizant.com
Certain concerns must first be addressed to take full advantage of big data (see Figure 3). They include: ●
●
●
●
●
Quantity of data. The amount of data and information available is gargantuan and only getting bigger, if projections are true. Business analysts and executives must decide what kind of data they need to use to address key business objectives. Right technology/tool. For better usability and results, business leaders need to select the most appropriate analytics tool. If the right tool is applied, it will help make sense of the data collected and streamline the solution to the business challenge. Association of structured, semi-structured and unstructured data. Regardless of its composition, data has to be used as a single, complete set and must translate bits buried in streams (e.g., keywords, speech, etc.) into machine-usable material. Specialist and talent management. To better exploit burgeoning data, organizations need a specialized talent pool that understands how all the information comes together and who can validate key insights. Big data calls for an approach previously applied to data mining but with a wider scope; the pool should consist of representatives from both business and IT, possibly using iterative development techniques and hypothesis testing, before scaling up. Speed of treatment/architecture. Given the vast amount of data and the diverse nature of unstructured information at hand, specialized platforms should be deployed alongside transitional solutions. At some point, the two will merge (e.g., combining unstructured social data conversations and transactional data). However, this will require a blended infrastructure in which the optimal enterprise information management architecture will support operational data collection in warehouses and a wide array of analytical tools.
The Four Axes of Big Data Velocity
Volume BIG
Variety
TA DA
Complexity
FIGURE 3 Source: Cognizant
Moving Forward with Big Data There are many grey areas that prevent organizations from leaping into enterprise-wide big data analysis strategies. By taking stock of the following points, organizations can get a better grasp on how to approach big data solutions.
http://cognizanti.cognizant.com
COGNIZANTi
7
●
●
●
Business leaders need to identify the desired value transformation objectives they want to achieve and the issues they want big data to solve. These issues can revolve around changing customer purchasing behaviors, pricing mechanisms, brand acceptance, service extensions, etc. (see use cases, page 82). A win-win model that benefits both customers and the business should result from this analysis. Once the business objectives are clear, an information value map (consisting of what, when, where, how) should be created. This should result in identifying the data needed to achieve these goals. Based on the value map, specific strategies/platforms/transformations/analytical processes must be defined; only then can strategic execution start.
The proliferation of data has augmented the need to have the right talent specially trained to realize ROI — both the return on information and investment. To effectively use big data, enterprises need to understand the parameters that are driving value in the minds of their customers and then gain an understanding into the various dimensions that must be analyzed. As discussed above, a variety of structured and unstructured data sources provide the basis for the insight needed to achieve this.
The Verdict: The Big Scale of Data Matters Big data is now being collected and sifted for insights that reveal hidden business secrets, ranging from customer preferences to performance drivers. If used properly, big data can help senior leaders make strategic decisions based on clearly delineated, customer-driven facts rather than gut instincts. The proliferation of data has augmented the need to have the right talent specially trained to realize ROI – both the return on information and investment. To fully succeed, senior leaders need to focus on how best to harness this data explosion and create use cases that build better and brighter enterprises. Big data has enormous potential to accelerate business growth through product and process innovation and increased employee and partner productivity that improve consumer satisfaction. Ultimately, the ability for any organization to survive and thrive in the marketplace depends on its ability to adapt and exploit enterprise and social data assets that can inform the radical change needed to overcome the inertia that prevails in most businesses.
8
COGNIZANTi
http://cognizanti.cognizant.com
Glossary Hadoop: An open source software framework for processing huge data sets on certain kinds of problems on a distributed system. MapReduce: A software framework introduced by Google for processing huge data sets on certain kinds of problems on a distributed system. Distributed system: Multiple computers, communicating through a network, used to solve a common computational problem. NoSQL Database: An open source database management system designed to handle huge amounts of data on a distributed system. Structured data: Data that resides in fixed fields, as in relational databases and spreadsheets. Unstructured data: Data that does not reside in fixed fields like free-form text, video, image and audio data. Semi-structured data: Data that does not conform to fixed fields but contains tags and other markers to separate data elements like XML, HTML or tagged text data. Visualization: Technologies used to create images, diagrams or animations to communicate the results of big data analysis.
Debasish “Dave” Mukherjee is an Associate Director in Cognizant’s Enterprise Information Management Practice. He has numerous years of experience in information management and consulting, managing and implementing strategic technologies in Fortune 500 corporations. Dave’s educational background includes MBA degrees in finance and MIS and a Ph.D. in information sciences from the University of North Texas in Denton, TX. He can be reached at
[email protected]. Karthik Krishnamurthy is the Global Business Leader in Cognizant’s Enterprise Information Management Practice. In this role, he drives strategic initiatives, business pursuits and key customer relationships across the globe. His focus areas include the creation of innovative next-generation solutions, entry into new markets and building thought leadership in the enterprise information management space. Karthik holds a diploma in business administration from IIM Lucknow and earned his bachelor’s degree from BITS Pilani. He can be reached at
[email protected].
http://cognizanti.cognizant.com
COGNIZANTi
9
