Why All the DM… Names?

Blog Post created by Advocate on May 6, 2016

The just-concluded FICO World 2016 was a great showcase for the Decision Management Suite, and the various product offerings and capabilities were well received. But we did get one consistent, and IMO justified, criticism—why does everything start with “DM” and what the hell is the difference between all of them? One attendee made the comment “there’s DMP, DMS, DMP-S; I feel like I’m in the DMZ!” While acknowledging that this state of affairs does indeed exist, and that it behooves FICO to clean up its naming, I think it’s worth looking at how this situation came to be. It focuses attention on both the strengths that FICO brings to the table and the challenges involved with promoting decision management technologies as standard development tools, rather than highly-specialized offerings for very select targets.


First of all, “decision management” is not a single activity or function. As viewed here at FICO, it’s a continuous process. We assume decisions are data-driven, and that new, relevant data is constantly being generated. That means that the business knowledge for making the decisions has to be constantly refreshed and informed by the new data. We’ve come up with various names for this: “learning loop,” “continuous improvement,” but whatever it’s called it’s critical to getting real value out of a decision solution. It also means there are a lot of moving parts. Marketing came up with a new graphic we rolled out at FICO World, that I think does a really good job of illustrating the continuous nature of decision management, and can help provide some context for where all those “DM” names fit into the picture.


You need to author decision logic. That can range from writing business rules to wrangling datasets to developing predictive or optimization models, and can involve numerous visual metaphors, algorithms, and development tools. Since this is your organization’s business knowledge we’re talking about (and which you just might have to show to some regulator), you need to be able to manage and track who wrote what, when, which version, and where it ended up getting used. To apply the decision logic it needs to be deployed as software services, and you need to be able to execute them in some data processing environment, which can be anything from a web service to a batch process to an ongoing data stream, with all the security, scalability, disaster recovery, and availability issues that need to be addressed. And then you need to capture, measure, and analyze the results, to learn whether or not you’re actually doing what you want, and to learn how to do it better.


FICO has been building variations on this cycle into its products for years, and that’s where both the strengths and the challenges come from. We understand the complexities and nuances, and we’ll put our resume and track record up against anybody’s. We also have a lot of ideas, design patterns, and technologies we can harvest and put into our decision management toolkit (i.e. the Decision Management Suite). The challenges come from figuring out how to package it all as a coherent, cohesive bundle, and how to best take advantage of the new and emerging technologies that are making the wide-spread adoption of decision management possible.


Our first thought was to put the word “decision” in front of pretty much everything. You might have also noticed that the words “model” and “manage” both start with “m,” and lots of our stuff is intended to either model or manage. Hence the plethora of “DM” names. Then we have the issue of emerging, and often overlapping, technologies. (Did you know that the Apache Software Foundation, which sponsors many of the most used and respected open-source software projects, has five different streaming data projects?) Choices made to address some customer needs and use cases end up being called one thing, while choices made to address other needs and use cases end up being called something else. The use cases may be valid, and the choices may be justified, but the differences involve those pesky nuances and complexities that are hard to translate into high-level blurbs and bullet points. If your cocktail-party conversations include lively banter about tuple processing and directed acyclic graphs, this might not seem like a problem, but believe it or not some folks don’t get excited about those topics.


So there you are. We have DM Everything All the Time. (D-MEAT? Maybe I should run that by Marketing!) Underneath all the initials, jargon, and buzz words is sixty years of experience getting real value out of real decision management solutions. Figuring out what to call all this stuff might take a little longer.


To learn more about the Decision Management Suite, see: