Ask 10 people how to define real-time analytics and you are likely to hear 10 different answers. Depending on one’s point of view, real-time analytics could, for example, refer to a key component of corporate strategy, a way of delivering custom offers through a Web site or a means of more closely monitoring business performance. On a technical level, it could mean executing an action based on a series of inputs or executing predefined data mining models in real time (for example, determining a customer’s propensity to accept an offer based on their history as well as answers to questions posed in real time by a call center representative).

 

Although points of view may differ, few would argue that the driver for real-time analytics is almost always the desire to become a higher performing, more competitive business. Studies such as those in Competing on Analytics have shown that businesses able to quickly turn insights into action are the highest performing among their peers – the companies that all others wish to emulate.1

 

The question then becomes, “How do I get started?” Some vendors present an all-or-nothing approach, while others an incremental one, but most make the mistake of emphasizing how to deploy complex predictive models to a particular operational system instead of addressing the larger issue of how to make smarter decisions in real time.

 

Introducing Decision Services

 

Decision services blends two concepts to help organizations implement real time analytics in a “right sized” fashion: 

  • “Decision” denotes an emphasis on selecting a given action for certain input, whether driven by analytics, business rules or some combination of the two.  
  • “Services” denote a focus on service-oriented architecture, a flexible way of adding powerful capabilities to existing systems, often by way of standards such as Web services.

Decision services make it easy for an organization to bring the benefits of real-time decisioning to its operations while allowing more sophisticated analytics to be added over time.

 

Decision services are applications within your application portfolio, or services in your service oriented architecture (SOA), that automate and manage highly targeted decisions that are part of your organization’s day-to-day operation. They are the tangible realization of adopting enterprise decision management (EDM), an approach that automates, improves and connects decisions to enhance business performance.

 

Decision services originate with a focus on business performance, not specific analytic technologies such as data mining or OLAP. This distinction is important, as real-time analytics are ultimately one of many tactics deployable via decision services. Pragmatic IT organizations know that real-time capabilities need not entail complex analytics. Decision services can act as a framework within which business rules and analytics coexist to achieve a desired business outcome.

 

Ideally, a decision service determines the most optimal choice each time it is called, one found using both rules and analytics and then chooses with known and acceptable levels of satisfaction and risk from available options. There is a clear growth path for decision services that builds up to this level, which helps justify such an investment as well as allows an organization to elevate its effectiveness over time.

 

Unfortunately, many early adopters of predictive analytical solutions have overlooked decision services in pursuit of real-time enablement, often for a particular customer interaction channel or line of business. Organizations yet to embrace predictive analytics in real time can learn a great deal from those who came before them.

           

Real-Time Analytics without Decision Services: Placing the Cart before the Horse

 

There is no denying the power of analytics to improve business performance. Predictive analytics, specifically technologies such as data mining, have transformed the way many organizations do business. Formerly impossible-to-imagine strategies, such as predicting customers’ actions and influencing their behavior by presenting custom products or services, are now proven among forward-thinking businesses.

 

The positive results achieved by these businesses, particularly those in retail and financial services, have spurred adoption of applications designed for capturing customer data, predicting behavior and pushing actions to interaction channels such as a call center or Web site in real-time.

 

These efforts were often driven by departmental or other siloed requirements and thus overlooked the implications of rolling out these capabilities more broadly. They also failed to address the resulting business requirements, such as resolving interactions among multiple channels, creation of mass customized products or services, or segmented marketing and offer management.

 

What happens is a waterfall effect of sorts, where new projects emerge aimed at transferring the value of the initial application to others within the enterprise. This approach can cause difficulty in managing and arbitrating among the many potential ways of interacting with customers – many of whom have benefited from predictive analytics through improved service and thus increasingly expect similar experiences when interacting with other facets of the business.

 

Quick fixes to these problems then result, offering short-term solutions that cannot adapt as requirements change. This situation begs for Decision Services that can centralize and automate the management of the many rules and analytics underpinning real-time, analytic solutions.

 

In the absence of decision services and the adoption of a distinct enterprise decision management approach, the decisions underpinning real-time analytic solutions tend to be embedded in the applications themselves and are typically aligned to a single operational system or business function. These lack the scale, flexibility and decision-centric analytic components offered via decision services.

 

Scale: As organizations become more customer-centric, they no longer view client relationships through a single lens, such as product. Instead, a multidimensional view, considering facets such as preferences, behavior and business objectives, lends a greater complexity to managing customer relationships. By exploiting a service-oriented approach, decision services offer high-performance decisioning, can accommodate growth in decision complexity and then adapted to multiple operational systems.

 

Flexibility: By maintaining decisions independent of applications and processes, decision services can be shared among other enterprise applications, including customer-facing as well as financial and planning systems. This not only allows sophisticated business strategies affecting multiple functions to be implemented quickly, it also ensures changes can be made just as fast.

 

Analytic components: Decision services also benefit from more advanced analytic capabilities for optimization. It is often not enough to automate a process with predetermined rules and predicted outcomes; there may be several “best cases” from which to choose, and a decision service can include optimization analytics to determine the best approach given business objectives.

 

As an enabler of sophisticated, high-impact business strategies, real-time predictive analytics can bring newfound capabilities to an organization - or have the opposite effect. When delivered as a robust decision service, predictive analytics can scale to the most complex environments and efficiently reflect the most current rules and policies of the enterprise.

 

Decision Services in Action

 

The following example of a global retailer is taken from Smart Enough Systems. Compared and contrasted are the two approaches used by the retailer to increase the value of its customers’ market baskets and drive sales across product categories. Note in the “EDM Way” example how both analytics and business rules combine to yield the highest possible return.

 

Old Way

 

This retailer designed its direct mail campaigns by hand. These particular campaigns were designed to get customers who had made purchases in only one product category to expand into a second or third product category by making an offer for a product in a category not previously purchased. Each campaign had creative elements combined with specific marketing offers, all derived judgmentally. The resulting letter was then sent to a large number of customers by extracting a subset of the customers using criteria designed to maximize the likelihood that the offer would be attractive. Letters were mailed, and an industry-standard response rate of less than one percent was achieved.

 

EDM Way

 

A decision service generates direct mail letters – it automates the decision of what letter content to use with each individual customer. Data from the data warehouse produces extensive predictive analytic models. These predict what products specific customers might buy and which products are particularly effective as first-in-category purchases. Business rules on marketing policies, objectives, target categories and more are combined with these analytics. The decision service generates a personalized letter for each customer. Each letter contains personalized information, specific recommendations driven by the analytics and information localized to the customer’s usual store using more business rules. The result is a completely focused piece of direct mail aimed at getting a specific customer to buy in an additional category.

 

Benefits

  • 2000 percent increase in response rates, to 17 percent,
  • Larger basket sizes for targeted customers, and
  • More purchases from new categories.

Results such as this are not uncommon for adopters of EDM and decision services. EDM allows an organization to begin reaping the benefits of more targeted and automated decisions via decision services, in this case with relatively straightforward direct mail marketing. Success in this offline channel can quickly be extended to real time interactions with customers via the web, call center or storefront. This growth path is a unique characteristic of Decision Services compared with other “all or nothing” approaches.

 

However you define real-time analytics for your business, remember to not only focus on the efficacy of the analytics but the larger issue of the framework in which the analytics are deployed throughout your organization. Abstraction has long been a programming paradigm but is fundamental to decision services, which isolate the logic behind your operational decisions, separating it from business processes and the operations of procedural application code. In an era of increasingly scarce IT resources, this has the dual benefit of simplifying the rollout of real-time analytical capabilities to various systems and greatly reducing the time, cost and risk of keeping up with changing business requirements.

 

Reference:

  1. Tom Davenport. Competing on Analytics: The New Science of Winning. Boston: Harvard Business School Press, 2007.

Register or login for access to this item and much more

All Information Management content is archived after seven days.

Community members receive:
  • All recent and archived articles
  • Conference offers and updates
  • A full menu of enewsletter options
  • Web seminars, white papers, ebooks

Don't have an account? Register for Free Unlimited Access