March 2004

By David Cameron, AptSoft Corporation

Imagine the following scenario: a customer walks through a department store and begins shopping, adding a couple of items that catch her eye to her shopping cart. Right before she checks out, however, she adds up the cost of the goods in her head and has second thoughts. Not able to justify the purchase, she walks over to an empty aisle, abandons the cart and leaves the store.

Meanwhile, a regular bank customer enters his local branch and picks up a credit card application, then moves to the counter and begins to fill it out. When he gets to the part of the form requesting his wife's social security number, he stops, realizing he doesn't have it handy. Deciding he will try to remember to get it and fill out the form the next time he is in the branch, he discreetly tears up the application and throws it away.
In the examples above, both the store and the bank were extremely close to 'converting' a prospect to a customer. They were on the cusp of finding someone who wanted what they were offering at a given point in time, had expressed enough interest to begin a transaction, but then were unable to complete it.
Let's continue the story.
At the department store, an alert customer service representative sees the customer abandon the shopping cart, recognizes her as a frequent shopper, and begins to reshelve the goods she left behind. The representative notices that one of the items in the shopping cart is about to go on sale. When she arrives for her shift the next day, the first thing she does is validate that the item is indeed on sale, then she looks up the customer's phone number in the phone book and calls her, letting her know one of the items she was interested in is on sale. The representative politely invites the customer back to make the purchase.
At the bank, a manager hears the application being shredded and looks up, recognizing one of her valuable customers. After he leaves, she reassembles the pieces of the application, noting that the spouse's social security number is the first field without information. Later that day, she calls her customer and invites him to finish filling out the form even without the social security number, suggesting that the application process can begin without it.
The lesson is obvious: 'closing' a transaction frequently requires a human being to address concerns, offer advice, prod and process. In the absence of this guiding hand, a significant percentage of transactions that rely completely on the customer will end unfulfilled.
Practically speaking, neither of these closing scenarios would take place in the offline world. Such scenarios are, however, very possible in the online world, where information is typically captured during the transaction. If the captured information had been analyzed and forwarded proactively to a customer service representative, a 'guiding hand' could have provided information to the prospective customer at just the right time, for just the right opportunities. Addressing this challenge represents a tremendous untapped source of revenue.

It's All In The Handoff
Despite the desire to automate many processes, many call centers still field calls from customers who could not complete online processes. Unfortunately, the customer often has to start over with the call center representative, wasting valuable call center time and creating the perception of bad customer service. Managing the 'handoff' from the online to the offline experience can better provide significant cost benefits by reducing call center time.
Making this hand-off smarter requires integration. However, integration's well-known cost and complexity have scared many organizations into thinking that making the online and offline world function seamlessly is greater than the perceived benefits.
New approaches to integration and technology, however, have changed the balance.
Traditionally, CIOs have had two ways to support the kind of requirements described above. First, they could write a lot of custom program code. However, code is extremely hard to change once written. That means that someone would have to query users to determine what they want the system to do prior to writing code. The cost for this type of approach is significant, which means that most CEOs will demand a detailed ROI with a 12-month payback schedule. Most of the cost of custom coding is due to two factors:
' Initial code must come very close to requirements, and requirements must be very specific, since it is expensive to go back and change after the fact; and,
' IT resources must be made available on an ongoing basis to troubleshoot, modify and upgrade.
The second approach is the data integration approach. By creating a common database, users and applications can share information and coordinate activity. This too, is very expensive, because it requires complex architecture efforts, as well as ongoing maintenance that involves refreshing data. Data integration's cost stems from multiple factors:
' Much of the cost is infrastructure to move and store data for potential future use (not known actual use) by application. This means most of the data are there for insurance purposes, and probably 20 percent of the data are accessed 80 percent of the time;
' Modifying business processes usually requires architectural rework, which drives up maintenance costs;
' The level of granularity of the data (i.e., do we need transaction level data or just summarized data of monthly transactions?) and the latency of data (i.e., how often do I need to refresh it?) drive key cost decisions ' decisions typically made data element by data element; and
' The cost of having IT staff hide the complexity of the physical data's appearance from the end-user who can't understand table layouts and other complexities also plays a role.
It is not surprising, then, that when faced with requirements to be more agile (to view change as necessary and good instead of costly and undesirable), the impracticality of these approaches rises to the surface.
Many end users, analysts and vendors have reached the same conclusion ' for the types of requirements emerging today in support of business imperatives, a new approach is necessary.
Integration And Automation With BPM
This new approach is business process management (BPM), and it relies heavily on new standards emerging in the form of extensible markup language (XML) and Web services to overcome the difficulties in prior generations of integration technology. The advent of these components enables a higher degree of automation and reusability than ever before.
BPM coordinates the actions of isolated IT systems ' both online and offline. Using the built-in intelligence of existing applications and systems, BPM can help companies nimbly react to key business events and capture millions of dollars in potentially missed revenue opportunities while improving customer service.
BPM technology overcomes the weaknesses of the current generation of systems integration approaches.
It provides a variable cost model that allows for a small initial implementation footprint tied to a three- to six-month payback period, tightly coupling investment and return. Today, technology must prove utility before vendors demand license and support fees. Instead of the typical 12-plus-month implementation and two-year payback model, BPM technology scales deployment effort to initial requirements. As a company extends its BPM implementation over time, it builds upon work already performed.
BPM allows for rapid modification so end users can engage in 'test-and-learn' process development and management. Most of the processes that support key business drivers are complex and must continuously evolve in response to internal and external changes. BPM technology enables users to implement, test, modify and re-implement processes in rapid sequence based on what they've learned.
BPM gives end users more involvement, so they can change processes on the fly and improve agility and productivity without having to call the IT department. The current generation of technologies is brittle largely because IT shoulders the bulk of maintenance costs and responsibilities. BPM technology balances the workload between the end user and IT in such a way that users can maintain processes without resorting to IT support.
Companies benefit from BPM's ability to reuse common data objects such as 'customer' and 'trade' and use open standards to make them available to any application. The complex infrastructure many organizations operate has evolved from many proprietary database structures and application logic syntaxes of years of systems development. Most valuable components of business processes, including data definitions, business rules and transformation logic, are replicated in a variety of formats across the enterprise. BPM technology relies on an object model that exposes and abstracts these elements so they may be reused across different systems via emerging XML standards.
BPM provides context for business processes to allow users to access information at the application level, before it has been saved to a database, dramatically reducing data integration. Most traditional application integration is based upon the movement of 'state data,' or data that have been saved about a particular event. State data are stored in databases, and then 'synchronized' with other databases linked to other applications. Typically, companies don't save much information needed to support business rules because it is too expensive or too complex.
Companies can use BPM to cultivate existing business logic and integration capabilities to connect existing applications and databases without modifying them. Most of these systems already have connections built in, either with middleware or published application programming interfaces (APIs). BPM technology uses these connections to link databases and applications while providing a layer of abstraction that hides variability between systems.
Finally, BPM automates many integration tasks that previously required additional code or manual configuration. For example, if a business process reaches a decision point and needs to know whether a given customer is 'high' or 'medium' value, programmers would typically need to manually code the 'fetching' of those data. Today's BPM technology can automatically provision information 'just in time' based on instructions provided by the user at the metadata layer. This allows the BPM server to automatically get that information without a user having to specifically say 'time to go get customer lifetime value.' This feature is especially powerful in dynamic environments where processes may change, resulting in a great deal of manual rework of data.
Gartner estimates that 75 to 80 percent of all integration projects are done with custom code. Despite integration technology's nearly decade-long evolution, it still is not a practical solution for most companies. As businesses evolve, brittleness must give way to agility in how systems support emerging business requirements. By taking a fresh look at integration from the process perspective, BPM technology has the potential to dramatically reduce the cost and effort involved with integration, generating tremendous benefits to both businesses and their customers.
Companies create competitive advantage by reducing the cost, time, complexity and technical skill required to connect applications. Many revenue-generating or cost-reducing business processes either fail or deliver marginal results because the underlying integration between supporting systems is weak or non-existent. In addition, even those processes that are automated are extremely brittle and difficult to change as market needs dictate. BPM can help organizations such as department stores and banks better use information to close the deal and seize revenue opportunities that once may have disappeared.

David Cameron is vice president of marketing and product integration at AptSoft Corporation (www.aptsoft.com) and can be e-mailed at david.cameron@aptsoft.com.

If you are interested in purchasing reprints of this article (in either print or HTML format), please visit Reprint Management Services online at www.reprintbuyer.com or contact a representative via e-mail at reprints@tmcnet.com or by phone at 800-290-5460.

For information and subscriptions, visit www.TMCnet.com or call 203-852-6800.

[Return To The March 2004 Table Of Contents]