Generic Business Intelligence (BI) tools and solutions for traditional applications such as reporting trends has been an essential tool for businesses to better understand issues such as the sources of business growth and act on those insights. The most potent of the class of BI solutions, predictive analytics, helps model and predict future outcomes that impact business performance in a form suitable for day-to-day operational use. Businesses with complexities of huge customer bases operating in highly competitive market environments will increasingly adopt advanced and specialized analytical solutions as a strategic weapon for growth.

Predictive Analytics

CXO�s of large and medium-sized businesses are increasingly recognizing the benefits of transforming their prized asset, customer data, into proprietary knowledge that can help the quality of operational decisions. Large customer base businesses such as retail, telecom and financial services in mature markets have invested significantly in strategic analytics programs to help drive revenues, market share and profitability. While vanilla BI reporting analyzes past business trends, predictive analytics provides a peek into the immediate future and a recipe for action. The new areas of predictive and behavioral analytics have now taken root in these businesses as an indispensable tool in areas such as developing preemptive and proactive customer marketing and service programs in highly competitive market environments.

The science of predictive analytics utilizes mathematical algorithms (such as those derived from life sciences, neural networks, etc.) and statistical methods to discover relationships in data (typically customer transactions) and make predictions. Innovative data mining technologies that allow results from these algorithms to be computed over terabytes of customer transaction data in �near real-time� form the technology basis for effective predictive analytics program. The last component is integration with tools for slice-and-dice analysis and reporting to provide an effective end-user platform for operational business decision support customized for functional areas such as marketing, sales and finance.

Use Cases

Advanced BI solutions such as predictive analytics are now integral to key decisions in several business functional areas across large and medium scale enterprises.

These include:
o Churn prediction: Be it predicting the risk of imminent churn of a mobile telecom customer or assessing the risk of lapsing to competition of a retail corporate customer, predictive algorithms allow effective customer retention strategies to be implemented. Predictive analytics based on behavioral models also provide insights on potential reasons a customer may be leaving in an manner that is easy to action.
o Target marketing: Predictive analytics can help a business identify its customers who have potentially a high affinity to purchase a specific product or service (existing or new).
o Demand estimation: For businesses that deal with sales of hundreds of products across large geographies, predictive models can provide knowledge key to sales planning: when is each product going to be purchased next, in what quantity and at what price?
o Fraud detection: Fraud detection algorithms help identify fraudulent behavior (e.g. credit card charges) in time for taking remedial measures.

Skills and Technologies

Predictive analytics demands unique and multi-faceted skills of personnel working in one integrated team environment. This includes mathematicians who can develop algorithms adapted to solving domain-specific business problems, domain experts in industry specific verticals such as retail, telecom and financial services who have knowledge of the respective business processes, and enterprise software product development experts to develop analytics products incorporating these algorithms customized for end users in the business.

Traditionally, due to the specialized skills-set required, predictive analytics capability has seen deployment through in-house captive analytics centers as in large financial services businesses. Companies that develop specialized statistical software, data warehouses or business reporting applications provide the tools but not the specialist mathematical expertise to solve a customized business modeling problem combined with the business domain expertise essential to map an effective analytics solution. This frequently results in watered-down analytical reporting (i.e. pretty graphs without clarity of insights) or complex statistical outputs that may not be meaningful for actionability. CXO�s of companies evaluating advanced analytics solutions in BI should be aware of these pitfalls to ensure a successful analytics program execution.

ROI

Unlike other technology intensive solutions, it is relatively straightforward to measure the effectiveness of predictive analytics as measured by such metrics as �prediction hit rate�, etc. and thus quantify their impact on incremental revenues or cost savings for the organization. Given the scale of their impact (e.g. across millions of customers), even a small percentage differential in their effectiveness results in a substantial impact on value derived. The higher investments necessary for deploying predictive analytics in BI is more than offset by the superior ROI.

�

Krishnamohan, Founder and CEO, Fifth C Solutions