IBM’s new decision management offering was formerly launched by Deepak Advani, VP Predictive Analytics, last month during a live webcast that also ran during the company’s annual Business Analytics analyst conference in Cambridge, MA. The event marks a significant point in the development of IBM’s decision management capabilities as it brings more technical cohesion to their offering – an offering that is designed to help automate and improve decision-making across a company’s core business systems.
Decision Management is a growing practice used to describe a set of business disciplines and technology for automating, optimising and governing repeatable business decisions. From a technology perspective it pulls together a range of components including business rules, predictive analytics and optimisation techniques. But this can also extend to business event processing, text, entity and cognitive analytics depending on the level of sophistication needed and what vendor you’re talking to!
With such a lot of technology falling under the Decision Management umbrella it’s no surprise to find that it has its roots in a number of different areas, and it’s also partly the reason why when a vendor, such as IBM, talks about Decision Management they need to position it from two different perspectives: operational decision management (ODM) and analytical decision management (ADM). This obviously begs the question: if decision management is all about managing operational business decisions, then why is there a need for both? And moreover what’s the difference?
It appears that part of the reason IBM talks about both operational and analytical decision management is that each tackles a slightly different variation on decision automation and improvement. Operational management for instance is applied where there’s a need to make the best possible decision at the current moment based on data and situational context, especially where decisions need to be completely automated through an ATM of self-service online application, for instance. Whereas analytical decision management is focused on using analytic data to discover insights that continually improve and automate decisions over time, in this case the output is used to support people, for example, via prompts to a call centre representative or a branch employee.
Of course an easier way for us (and IBM) to differentiate between ODM and ADM is from a technology perspective and how different technology components can be applied to support each approach. As IBM sees it, operational decision management (ODM) brings together a business rule management system (BRMS) and business event processing and is applied to decision-making situations where lots of rules are required and there’s a high degree of certainty based on things you know, such as corporate policies, human knowledge and known events. Here business event processing identifies patterns in the data and the BRMS applies reasoning logic to make the appropriate decision based on the context of the situation. On the other hand, analytical decision management (ADM) combines and applies predictive analytics, basic rules and optimisation techniques to situations where there’s a high degree of uncertainty. This might include things like what’s most likely to happen in the future or what is the best solution based on the interpretation of data trends, by using analytics to uncover insights in the data to help improve decision-making.
Much of this separation in approach appears to stem from the different places each product offering has come from historically, rather than any big fundamental difference in goal or rationale. However despite this it’s not always clear to me how you might draw the line between an ODM or ADM approach. Surely for example, applying decision management to help identify and take action on suspected fraud threats should be an ADM endeavour? Especially because there’s a high degree of uncertainty to the situation and there’s also a need to leverage analytics to predict ’things’ that have a high probability of being fraudulent. But that’s not always the case as IBM sees it; there are also scenarios in which you may want to tap into the event processing and BRMS technology from ODM too, to help identity or respond to fraud in real time, especially where you are processing high volumes of credit card transactions for instance. Such a scenario requires a combined or integrated approach that incorporates bits of both ODM and ADM – further blurring the boundaries between both approaches.
I suppose the bottom line is that any decision (excuse the pun) about ODM or ADM shouldn’t only be predicated on technology requirements, but also needs to be grounded in the business opportunity or challenge that it aims to solve such as:
- How can I streamline my claims payment process to improve efficiency and reduce claims costs?
- How can I increase the revenue and profitability of my cross-sell and up-sell activities?
- How can I make the right next-best-offers based around my current marketing budget?
In this respect I do believe IBM needs to be a lot clearer about the use cases for ODM and ADM – and where you might want to use both. Ideally this clarity would help remove some of the technological distinction between both approaches altogether.
But what do you think? Can you tell the difference?
As an aside, we wrote about IBM’s Analytic Decision Management offering last year in a Vendor Insight report. Although it’s not focused on the latest version of ADM it does provide a thorough overview of the offering and how it works in the context of improving customer interactions. If you want, you’re welcome to download it for free here*.
* A note from the blog editor: Please note that this report was funded and published independently by MWD Advisors with no editorial input from IBM. They simply valued Helena’s comprehensive analysis enough to want to purchase the reprint rights. You can also see the report in the context of our own research library here. Read more about our research approach and principles.