BPM Strategies - March 2008 - (Page 17) Taylor: Straight-through processing is one of the key reasons for doing this. All too often when people talk about straight-through processing, they talk as if just linking the systems will be enough. But in most of those systems, transactions must wait for people. While it is clearly useful to use a BPM approach to link the systems and move the data correctly, and to tell somebody who uses system B that system A is finished can make a big difference, it doesn’t get you to lights-out or straight-through processing. technologies that are important? Taylor: Business rules management has proven itself to the point where you can manage a lot of rules and deploy them easily to a services-oriented architecture so they can be accessed by your BPM tool. This is a core capability. The other technology is data mining and predictive analytics. Those are still by and large aimed at a highend user. They are not business-userfriendly by any means. But companies have a lot more data; it is cleaner and more accurate than it used to be; they understand it better because they have invested in BI, so they are able to ask the right question more often. The tools and technologies that exist for doing data mining and analytics have gotten much better at deploying the results - either generating business rules or standards-based output like PMML, the Predictive Model Markup Language, or just generating code to execute. It is a lot more practical to bring them into production applications. you and the card that go into deciding whether it is fraudulent. Where would I display that information and to whom? So I don’t. Instead, in the back, there is a sophisticated set of rules and a neural network that determine how likely a transaction is to be fraudulent. The rules balance what the analytics say with the card issuer’s best practices and procedures. BPMInstitute.org: So those are the technology drivers? What are the conceptual enablers? Taylor: The first is an explicit focus on decision management. You often hear people talk about business rules and analytics in the context of a business process, as though it were all the same. But that is not exactly the case. Just like business processes are not the same as systems, business decisions are not the same as business processes and they should be managed differently. BPMInstitute.org: How do you get to “lights-out” processing? Taylor: You need to be able to make intelligent decisions when no one is there. You want to say that for, say, 95 percent of the transactions, you can automate the decisions and in the morning a manager can review the five percent that are tricky. Business decision management injects that intelligence into the business process management framework so you can get handsfree, lights-out, straight-through processing. The data does not stop and wait for anybody to okay it. The system is smart enough to take decisions and flow most of the transactions straight through. BPMInstitute.org: How close are we to being able to apply data mining and analytics to a business process? Taylor: You are beginning to see a realization that traditional BI tools can’t simply be applied to a real-time application. It is not just a question of getting the performance. If you are trying to do straight-through processing, a report isn’t very helpful. What data mining and operational analytics do is see what a smart person would see if they could visualize the data and turn it into something that can be executed - a formula or a calculation or a set of rules. Because the end result of data mining and analytics is executable, it is a better way of turning your data into operational insight. You want to be able to help the system to make a decision. BPMInstitute.org: How are they different and how should they be managed differently? Taylor: A lot of the things that are prioritized when you are managing a business process such as state and flow are often less important in a decision-sense. Fundamentally, most businesses have processes and systems that reflect the way that they do business today. They could completely change the way they do business today and it would not fundamentally affect the business decisions they have to make. Let’s say I want to know how aggressive a retention offer to make. That is based on a bunch of different things. It doesn’t matter if you interact with me by using my call center or my online interface or go to my store or a third-party store. That all has to be tied into the systems you use and the business process you use, but it doesn’t affect the decision you make. Managing decisions as a corporate asset and not just as part of a process makes a big difference. BPMInstitute.org: What are the enabling technologies that make thinking about this reasonable? Taylor: There are five things. A couple are conceptual and some are technology-driven. From a technology perspective, there are three main things. First, the service-oriented architecture/business process management environment - the idea that you are assembling applications from components or services with different properties. If you identify decision services as something different - which you need to do to make it work - it is easier to do if you are already thinking about different services doing different things. The mindset that you are not building an application but you are building a bunch of services and are assembling an application makes it easier to think about decision management. BPMInstitute.org: So you need analytics and the rules that tell you what to do with the results? Taylor: Very much so. The rules and analytics in combination make a big difference. If you think about credit card fraud, when you swipe your card, there is a lot of data about BPMInstitute.org: Are there specific March 2008 BPMInstitute.org: Decisions should be viewed independently from processes. 17 http://BPMInstitute.org http://BPMInstitute.org http://BPMInstitute.org http://BPMInstitute.org http://BPMInstitute.org http://BPMInstitute.org http://BPMInstitute.org http://BPMInstitute.org
For optimal viewing of this digital publication, please enable JavaScript and then refresh the page. If you would like to try to load the digital publication without using Flash Player detection, please click here.