Ronan Bradley had an interesting article on ebizQ this week - Fertile Ground for ROI in BPM: Three Unlikely Areas. In it he outlined some areas of banking where business process management (BPM) could deliver an ROI.

1. Keeping up with regulations
   In which he points out that “a feature of BPM systems (over custom coded solutions for instance) is that they are configurable through the use of rules”
2. Derivative Trade Cycle
3. Automation of back-office and support desk function
   In which he says that “All of this makes a good potential sweet spot for BPM: complex sets of rules to be followed and automated, and high cost associated with human error”

What struck me about this, and should strike you, is how decision-centric Ronan’s examples are. Regulation is often about what is allowed or not allowed - what decisions are appropriate in other words. Part of the derivative trade cycle is pricing - making an appropriate pricing decision - and part is about eligibility decisions. Automation of the back-office involves automating lots of know-how - the rules behind decisions.

Now I’ll be the first to admit I have a decision-centric point of view but I do find it interesting that three areas of ROI for BPM could have so much to do with decision automation and management.

ShareThis

My good friends Larry Goldberg and Barbara von Halle have just published an interesting article: The New Frontier: BPM, BDM and SOA. It’s worth a read as it makes some good points about the intersection of BPM and decision management - whether Business Decision Management or Enterprise Decision Management.

ShareThis

Ed Garry of Oracle wrote a piece for Wall Street and Technology called Analytics Help Firms Turn Data Into Opportunity that I found last week. In it Ed talks about Real Time Decisioning platforms that “deliver both rules and predictive analytics to power solutions for real-time enterprise decision management”. Ed is, of course, correct though I would add adaptive control to the list of core capabilities needed for real time EDM.

Ed makes lots of good points in the article although he says one thing that might surprise readers of this blog. He says “A central component of enterprise decision management is communicating with sales targets based on their preferences”. Well, clearly enterprise decision management is not only about sales and marketing but even when you consider EDM only in that context it might seem strange to describe preferences as a central component. But think about it for a moment and it makes sense - preferences on the part of a customer or prospect are rules about how they want to be treated. Managing these rules along with the rules that are driven by regulation or policy or mined from your data makes perfect sense. While I am not sure I would describe it as a “central” component it is clearly important as I have said before when discussing personalization.

ShareThis

The folks at ILOG and Relativitiy recently announced a new integration between their products - Legacy IT Modernization enabled by ILOG and Relativity Technologies Business Rules Solutions. I got a chance to chat with them today about what was new and different in this latest attempt to bring legacy modernization and business rules together. Relativity’s product line allows a variety of languages (COBOL, Java, PL/1) to be analyzed and the resulting model of the application is stored in a repository. From this they make the information available to developers maintaining or redeveloping code, managers making portfolio decisions etc. Business Rules Manager is one of the products that interacts with the repository and is designed to mine candidate business rules from the model in the repository with an end goal to evolve them to the point of being useful in authoring new rules. In particular their product, they say, helps move the very technical “rules” that come out of code analysis to something more useful. They can differentiate between real business logic and what you might call housekeeping code. The repository allows you to build a business vocabulary that can be assigned to the structures found in the code and thus to the rules.

The initial integration with ILOG’s product takes this vocabulary across as well as copybooks etc. This would allow new rules to be authored using the same object model and vocabulary as that established in the model in Relativity. At present the candidate rules remain in the Relativity environment and are not moved into the ILOG repository but these candidate rules and some elements of ruleflow are next on the integration roadmap (though no date has been announced yet). In addition Relativity is working on decision tables and decision trees and the companies expect this to help with integration as many trees and tables should come across more or less wholesale. The current integration is with JRule and takes advantage of some of the features in Rules for COBOL (which I blogged about before).

When companies reengineer legacy applications into COBOL one of the most important aspects is impact analysis. Often some of the legacy code is left untouched, some is restructured and some is completely replaced. Ensuring that impact analysis is done right across all of these is often a challenge. The folks at Relativity re register the COBOL generated from ILOG’s product. This code and is structure, comments etc is then in the repository and this code contains information that links to rules artifacts. This allows impact analysis to be done in the Relativity product that will include the impact on code generated from the rules and, indirectly therefore, to the rules themselves.

This is clearly a first version of the integration as there are still a number of areas where no work has been done. Besides the obvious lack of integration for candidate rules, there is also no work so far on how a user of the rules environment would be able to access artifacts managed in Relativity to help them write the rules correctly. Similarly no work so far on rule templates. This is important when the legacy code is being replaced with rules because it changes so often. In those circumstances the current values of the rules may be less interesting than the structure of the rules in the code. Taking the structure of the old rules and creating rule templates from them may be more useful than just moving the rules themselves.

ShareThis

Transpromotional marketing - yes, another new phrase that I heard for the first time this week. Wooing Customers in a Weak Economy was the source - an article on 1:1. Chris Stone wrote the article and it talks about the need to use different channels to contact customers and to do so consistently and in a personalized fashion - both hallmarks of using decision management of course. The phrase “transpromotional” refers to the use of transactional opportunities, such as statements, to deliver personlized promotions and marketing.

Taking this approach requires both an act of will - deciding to use these channels to deliver personalized marketing - and then a focus on the decisions involved. What offer/action is appropriate for this customer? How can I personalize it and make it attractive to them? How do I correctly include risk in my calculations as to what to offer. All of these decisions are classic opportunities for decision management or EDM as they take rules (policies, procedures, customer preferences), analytics (segmentation rules mined from data, predictions of risk and retention as well as opportuinty) and constant testing and refinement or adaptive control.

Transpromotional marketing - another use of EDM.

ShareThis

Ellen Goodman of the Boston Globe had a column “Self-serve and slave” (that I saw in the San Jose Mercury News as “In a self-serve nation, work gets dumped on us“) in which she rails against self-service and compares it to the outsourcing of work from paid employees to us consumers. As she says:

For every task shipped abroad by a corporation, isn’t there another one sloughed off onto that domestic loser, the consumer?

This perspective was interesting. Speaking personally I often prefer self-service - I don’t like to have to wait for someone to serve me in a shop and I find it irritating to talk to someone who is using an information system rather than using the system myself. Indeed I find many circumstances to be much more efficient when I can self-serve. I suppose that an infinite number of staff available 24×7 might work as well as an ATM say but well designed self-service can make machines (PCs, kiosks, ATMs) do something that otherwise a person would do. Each such automated channel then adds to the number of available staff (typically finite) and so makes it easier for everyone. For a good example, take the check in kiosks at airports. Where there are plenty of kiosks the lines for staff are shorter and I think most people are happier.

I guess you could consider me as a glass half-full person when it comes to self service while Ellen might be described as a half-empty person. So, this week’s question is “What do you think - is the self-service glass half-full or half-empty?”

Regardless of the perspective you have on this I believe that Enterprise Decision Management has a role to play. Being explicit about the decisions that customers want made makes it more likely that they can get done what they want done (because it is explicitly automated). Because personalization makes it easier and more pleasant to use automated systems, explicitly identifying the decisions that can be personalized i also helpful. Lastly it helps as it allows the expertise necessary to make a good decision to be embedded so you are not making decisions based on TV shows and google searches or the uninformed new hire on the other end of the phone.

ShareThis

Gary Cokins posted How are Enterprise Decision Management (EDM) and Performance Management Different? in response to my original post What’s the difference between EDM and Performance Management. Gary takes me to task for a narrow perspective on performance management and, to some extent, he is right. Too often what I see described as performance management is just performance monitoring. Dashboards and graphics that show someone how well they are doing at meeting some KPIs, perhaps presented as a balanced scorecard or similar. There is no way for someone looking at these dials and graphs to do anything about them - they must leave their dashboard and go off and tell someone to change something. They are dashboards, not cockpits, as I said before.

Gary’s more complete and richer view of performance management is a much better one than the typical implementation. If you took his approach and really thought through how someone managing the performance of their organization against their objectives could take action to change things - to improve performance as he says - then there is to my mind still a gap. Here’s a quote from Gary:

[Performance Management links an] executive team’s strategy formulation to operations for strategy execution. It gives context to its purpose

Even when this is true I believe it ends at the systems portfolio. Most executives and line managers have no way to change the behavior of their systems. To the extent that the execution of strategy is automated (and this is more and more true in today’s high speed, 24×7 world), most performance management implementations do not really link to operations. This is where EDM comes in. Ensuring that the results of individual, operational decisions can be measured and improved by the business people who understand the objectives even when those decisions are embedded in processes and systems that are wholly or partly automated takes EDM.

Performance management may give a context to EDM but EDM ensures that performance management does not stop with people but extends to the behavior of your systems also.

ShareThis

Thanks to my friends at Bankwatch I heard today about eBank in Japan. The bank is described in this nice article on swiftcommunity.net. What struck me about this was the focus on complete automation of decision making - not just the processes and information colleciton, but the decision making itself. Is this the person they claim to be? Should we allow this person to open an account? I bet they also handle things like refunding fees, granting credit and much more in a completely automated way also.

I think this is the future of banking not only because mobile banking is exploding and absolutely requires this degree of automation but also because I believe that this kind of automation let’s staff in branches be more productive and more customer-focused and because it makes it possible for a “branch” to be a person with a computer - in a grocery store, in a mall, at an event. Taking banking to the people not people to the bank.

Let me know what you think.

By the way, we passed 500 subscribers for the first time today so thanks to all of you who make the Smart (enough) Systems blog part of your regular routine! If you like the posts and are not subscribed already, you can use RSS (http://feeds.feedburner.com/smartenoughsystems) to add the blog to your reader or you can click here and subscribe to the email version.

ShareThis

Version 17? Yes, SPSS has been at this a while. Today they announced version 17 of their Statistics package. While this has some new features to handle more data and some improved asset management (an analytic repository), the big features are really about bringing more business users into the analytic fold. One feature in this area was a new interface to allow marketing professionals to do Recency, Frequency, Monetary (RFM) analysis - a very common marketing analysis. Another allows collaboration between technical and non-technical users by allowing technical users to create business-friendly interfaces for their users. This allows business users to do their own analysis on common tasks using custom-designed interfaces built right into the tool and reducing the learning curve.

For power users they have added a syntax editor for SPSS Statistics programmers to make programming SPSS Statistics faster and easier. Integration with Clementine and Dimensions to share data access, analysis and reports makes this release more integrated into analytic processes across the enterprise while mulitple imputation for missing values (always a problem in analytics) and better integration with Microsoft Office and Adobe Acrobat round out the release.

ShareThis

A: www.youtube.com/watch?v=1zDwIfSDQiE

(Falcon, by the way, is the name of Fair Isaac’s fraud detection product that combines a neural network - a form of predictive analytics - with business rules and dynamic profiles to make real-time decisions about credit card fraud).

ShareThis