Advanced Analytics The Way Forward for Businesses Dr. Sujatha R Upadhyaya Nov 2009
Advanced Analytics Adding Value to Every Business In this tough and competitive market, businesses are fighting to gain new market share and retain old customers. They need business intelligence that can predict future scenarios, identify opportunities, anticipate problems and improve processes. The right Advanced Analytics solution can effectively meet these expectations by providing valuable insight through analysis of structured and unstructured data. An ideal solution should not only be designed from a business users' perspective but also be versatile, easy to use and capable of presenting findings in an intuitive manner. 2 Infosys White Paper
Heralding a New Phase in Analytics Today's economic scenario is characterized by stiff competition and demanding customers. Businesses are striving hard to achieve their objectives of retaining customers and raising profitability by providing support to decision making. By leveraging Advanced Analytics, that goes beyond the basic Business Intelligence (BI) functions such as querying, analysis and reporting, one can make informed decisions using real-time data to gain competitive advantage. Advanced Analytics spans across customer analytics, predictive analytics, text analytics, decision support and optimization applications and represents an evolution in BI, where its true potential has been leveraged by developing insights that are not directly evident in historical data. According to a global research firm, despite the economic downturn, Advanced Analytics revenue increased to US$ 1,522 million in 2008, at a growth rate of 12.1%, a clear indicator of increasing adoption of advanced analytics solutions. Another survey conducted by a top research company early last year indicated that of all emerging technologies, Business Intelligence and Analytics would have the greatest impact on business in the next two to five years. Infosys White Paper 3
Using Unstructured Data to Create Actionable Insight An exciting new trend in analytics is the use of unstructured or informal information as input, in addition to structured data. While a considerable amount of structured data - transactional, demographic, etc., is mined to create intelligence, an enormous amount unstructured data lies unused. An ideal Advanced Analytics solution should have the capability to extract meaning from the highly voluminous unstructured information and convert that into actionable insight. Unstructured data comes in a variety of forms and modes including, call centre narratives, blog feedback, wiki entries, customer service notes, tweets and email texts. Mining such information is quite complex and requires more than analytical techniques. The answer lies in an innovative technique called "Natural Language Processing", which is capable of processing human language into semantic insight. A tool with such capability can be configured to explore a host of online and offline unstructu red data sources, pick up the sentiments expressed for further analysis and present the findings in a matter that is easy to understand and act upon. The technology has already found acceptance in businesses that are heavy on customer interaction such as telecom, financial services and retail. Besides text and verbalised inputs, it is also possible to analyze images such as video files and photographs. A couple of applications include - the retrieval and study of patient x-rays, scans and other diagnostic images stored within a hospital's database to make future diagnoses; and the analysis of traffic videos at busy intersections to predict vehicular movement. 4 Infosys White Paper
An Advanced Analytics solution can be a valuable tool at the hands of key stakeholders in the decision making process viz. business managers, analysts. BI experts and, other decision makers such as CXOs and Analytics/BI heads. It finds application virtually in every business area including forecasting, campaign management, competitor analysis, customer retention, portfolio management and more. Advanced Analytics comprises techniques to analyse and model data, giving enterprises the necessary insight in order to: > Optimize processes > Identify opportunities > Anticipate unfortunate turns > Understand customer behavior and demand patterns Users across functional areas such as sales and marketing, human resources, finance etc., can leverage Advanced Analytics to bring about: > Improved customer relationship as a result of better servicing and targeted cross-selling > Better understanding of the market by way of competitor, customer and sales analysis > By detecting opportunities early, companies chance upon setting trends rather than following them > Better clarity on product and field marketing initiatives > Value addition to Research and Development activities > Better management of human capital and resources > Higher efficiency in financial management > Reduced risk as a result of better anticipation of negative events Advanced Analytics Benefiting Every Type of Business User Infosys White Paper 5
Advanced Analytics Platform for Supporting Multiple Applications Given the diverse business needs addressed by Advanced Analytics, there is an immense potential for an Advanced Analytics platform; one that can support quick development and deployment of advance analytics solutions for individual needs. The platform should have the ability to develop, deploy and manage multiple algorithms and different models which encode the intelligence behind the making of predictions. Advance Analytics solutions that stem out of this platform include Customer Analytics solutions for recommendation and customer profiling, Predictive Analytics solutions for predicting customer behaviour, customer churn analysis, and channel preference and also, text analytics solutions for improving the predictability of other analytics solutions by bringing in information from text data. Voice of Customer solution is a typical text analytics solution, where customer feedback in e-mails, blogs, complaints, discussion forums are processed to give better insight into customer pain points. 6 Infosys White Paper
Advanced Analytics applications today are often known by the functional area they is being applied to, such as Claims Analytics, Financial Analytics, Stores Analytics, etc. An ideal Advanced Analytics platform must be able to support the development of solutions in all of these functional areas. CLAIMS Claims Analytics has gained pace in recent years as a tool to help businesses understand the methods to improve business processes from analysis of claims related data, thereby enhancing customer experience. The insurance industry typically has a huge amount of claim-related data at its disposal. Until recently, most of this information lay unexploited, since it originated in unstructured form. With text analytics capability, one can get valuable insight into fraud, delayed settlements and other problem issues. The insurance companies can then take remedial action by assigning people with the right skills to handle tricky cases and optimizing their processes to ensure smoother customer experience. FINANCIAL Financial Analytics helps in enhancing business performance in the financial services domain through services such as budgetary analysis, forecasting, working capital management and shareholder metric analysis. Banks and insurance firms are increasingly using financial risk analysis, as a part of Financial Analytics, for minimizing risk associated with lending and issuance of cover and for protecting against fraud as well. STORES Stores Analytics addresses the need for efficient management of store space and inventory within the retail industry. It enables stores to overcome bandwidth constraints by triggering activities through realtime alerts. Store Analytics has other interesting features which enable store layout analysis, shrink analysis, shelf management and cross-selling promotions. Using Stores Analytics, retailers are able to improve operational efficiency, identify fraud, resolve compliance issues and earn a rapid Return on Investment (ROI). CAMPAIGN Campaign Analytics improves the effectiveness of campaigns by providing businesses with important information about market segments, customers, channels, etc. It also helps in quantifying the success of campaigns through measures such as ROI and conversion influence. MOBILE Mobile Analytics enables mobile site owners to gain insight into their site traffic and usage. Mobile Analytics is functionally similar to Web Analytics, but its scope has been adapted for mobile devices. As mobile devices have become increasingly ubiquitous, one can expect much traction in this space in coming years. COLLABORATIVE Collaborative Analytics is an emerging practice that enables more than two stakeholders to combine their efforts to achieve common objectives instead of carrying analytics as an isolated activity. In short, the goal of Collaborative Analytics is to generate wisdom through collective effort. As the name suggests, such analytics relies on collaboration to collate data, verify and share results and co-ordinate action. Infosys White Paper 7
Key Expectations from Advanced Analytics Solutions Thus, a versatile Advanced Analytics solution would enable mining of both structured and unstructured inputs to create intelligence for businesses. While evaluating competing analytics solutions, enterprises may employ the following assessment criteria: Ease of Use: Analytics is no longer the preserve of statistical experts. Today, the analytics tools are being used by business users as well as decision makers. Therefore these tools must be simple and intuitive to use and capable of presenting findings in a friendly and understandable manner. In this context, the quality of dashboards used for presenting the findings becomes very important. Accuracy of Analysis: Accuracy of prediction or analysis is a vital factor associated with acceptance of an Advanced Analytics solution. It is extremely important to choose the right algorithms / approaches in the right context. Many algorithms that seem to work fine in simulated environment fail when it comes to perform in actual situations. As almost all of these solutions involve incorporating a great deal of domain knowledge, domain expert involvement in model building phase is extremely important. 8 Infosys White Paper
Low Total Cost Of Ownership (TCO):For justifying an investment in an analytics solution, it must come at a reasonable cost of acquisition, implementation and maintenance. Typically, the platforms for deployment of these solutions are built with wide range of capabilities. However, depending on the need, it makes sense to acquire solutions that include only those capabilities appropriate to the context instead of the complete capability of the platform. Although, this might restrict the user from developing new solutions, it helps in reducing the cost considerably. Vendors have also started to offer components or even entire solutions 'on demand' in order to reduce the investments that their customers need to make upfront. Quick Deployment: In a dynamic and cost conscious environment, businesses can ill-afford long lead times for project implementation. This applies to the implementation of any Advanced Analytics solution as well. One of the innovative ways to speed up project implementation is through the creation of industry-specific packaged solutions that are nearly 'ready-to-deploy.' Using platforms that support multiple built-in algorithms and models to enable quick development of contextualized solutions also play a part in speeding up deployment in this context. Ability to Analyze Unstructured Information: An Advanced Analytics solution that can also process unstructured data is more likely to provide better predictions. This often forms the differentiating feature for an ideal solution as this feature is not ubiquitous in the marketplace. Compatibility with Existing IT Infrastructure and Ability to Integrate with other Applications: It is imperative that the analytics solution be seamlessly integrated with the other enterprise applications that will provide data feeds for analysis. Apart from being able to retrieve data from multiple sources, the solutions must also be capable of preparing it for onward processing. This means cleaning, de-duplicating and verifying data that is available and logically filling up any gaps that might exist. Scalability: The analytics solution must be capable of keeping pace with business growth for years to come. It is vital that the solution be sufficiently scalable to handle increases in data input as well as number of end users, without falling behind on performance or reliability. It would be good to use an open architecture to make provision for adding a new algorithm or building a new model. Infosys White Paper 9
Summary The capability of Business Intelligence has extended beyond querying, reporting and data analysis and encompasses the entire gamut of Advanced Analytics that includes the latest trends in analytics. Businesses are showing increasing affinity toward adopting Advanced Analytics to manage customer relationships, understand market movements, anticipate the fallout of technological developments, defend market share, etc. New functional areas such as Financial Analytics, Claims Analytics and Stores Analytics that have emerged in this space to cater to the needs of specific verticals such as Finance, Insurance and Retail are addressed by Financial Analytics, Claims Analytics, Store Analytics correspondingly. Regular analytics tools work with structured data. But, the ones that can process unstructured information can bring in insights from textual content as well. This adds a keener edge to decision making. Finally, although several analytics packages dot the landscape, only a few promise ease of use, quick deployment and a quick return on investment. Before acquiring a particular solution, businesses must be sure of the capability of the solution to deliver these key benefits. REFERENCE Beyond Spreadsheets: The Value of Business Intelligence and Analytics, January 2009, Aberdeen Group Market Outlook for Text Analytics, by Seth Grimes, April 23, 2009 About the Author Dr. Sujatha R Upadhyaya is a Researcher with Information Management group in SETLabs Infosys. In her present assignment, she focuses on leveraging intellectual properties in the space of Information Management for building innovative solutions that address industry needs. Her research interests include knowledge modeling, data mining, text mining, machine learning and their applications to analytics solutions. She has authored several peer reviewed research articles in International Conferences and Journals. Her experience spans across the roles of an academician, a researcher, a data and text mining expert and an analyst.