One of the principals and author of Decision Management Systems(IBM Press) and The MicroGuide to Process and Decision Modeling in BPMN/DMN, James Taylor joined me for an overview of the subject.
Download PDF Transcript of Podcast
Note: This is a transcription of a podcast. It has not gone through a professional editing process and may contain grammatical errors or incorrect formatting.
Related Podcast: The Systems and Processes of Decision Management
Transcription of the Podcast
Joe: Welcome everyone! This is Joe Dager, the host of the Business901 Podcast. With me today is James Taylor, not the singer but the business process modeler. He is one of the foremost thinkers and writers of Decision Management with over 20 years’ experience in developing software. His latest book is Process and Decision Modeling. Thanks for joining me James and I know you’re not in the country right now, where do you happen to be?
James: Right now Joe, I happen to be in India. I’ve just been out training a bunch of folks in decision modeling, interestingly enough and so I’m just about to head back to California.
Joe: Can you give me just the elevator speech, your company name and where you’re located.
James: Sure! I’m CEO of Decision Management Solutions. We’re a consulting firm specializing in decision management and decision modeling. So, how do we use technology and modeling techniques to embed decision making in our system’s portfolio. How do we take the decisions that we have to make as part of our day to day business and our day to day processes and automate those decisions effectively using everything from business intelligence to business rules, predictive analytics, and deliver improved business results as a result?
We’re based in California, though we have folks in various places across the country like many virtual organizations these days. I’ve been in this space in decision management for over a decade and doing decision modeling for the last three or four years probably.
Joe: Well then you’ve probably been in the space before there was software to do it?
James: Well for decision modeling, for sure. I mean, the decision management space builds on predictive analytics, and risk scoring and business rules, and technologies that have been around awhile and I have a long history of working together. But decision modeling is more recently. It’s been something that a number of communities have been working on, and it has really started to come together with standards and software support really only the last year or two.
Joe: Explain to me, most of my people are Lean people and service-design type people that listen to the Podcast and this seems kind of like a step from our regular customer journeys and our process mapping. Is there a definition or a distinction that you can make for people?
James: Familiarization is tightly linked, right? What we find is that if you look at a process map or a customer journey within that process, within that journey, there are many points where you have to make choices. You have to make decisions about what you’re going to do the next. What offer am I going to present? Can I refund something? How do I price this? Is this person eligible for something? And all these decisions points, what we find is one of two things happen; they both get glossed over and not really described or they get handed off to a person to make, we just punt as we say. Well, we don’t know how to do this. We just ask people who are in the realm to go do it for us.
A few people kind of model with their process models, but process models don’t do a good job of expressing how we want to make a decision. What we see is people are starting to say, “Well look, I’m going to get my journey, my process map to the point where I’ve got to make the next decision.” Then I’m going to say, “Okay, that’s what I got to do at this point and I’m going to build a model of that decision that would tell me how I’m going to make that decision, what information I need to make that decision, where I’m going to get the rules, the knowledge I need to make that decisions, and model that out so that I can either be clear how I want to make it manually.” Or I can start and say, okay, this piece of this decision, maybe all of the decisions, maybe some of the decisions some of the time could, in fact, be automated by my systems portfolio. I’m going to build a decision management system and not a decision support system that helps a person make the decision, but an actual decision management system that makes the decision for me makes the recommendation, makes the choice, and then embed that to get more straight through processing, more precise, targeted customer treatment, whatever it might be based on a very robust model that I needed to do.
Joe: When I’m looking at this, this is going to bring forth things that maybe I didn’t really consider before or bring data in that I’m not looking at, so I’m going to be able to make a better, informed decision?
James: Yes, that’s certainly one of the usual cases. One thing, we find, is that when people start to build these models, then they start to actually say, “Well, how do we make that?” We like to use questions a lot, so we often say, “Well, what’s the question you have to answer to make this decision?” You define that and then you say, “Well, one of the questions you have to answer before you can answer that question…” and you can sort of drill down in that sort of reflective way and what you find out is that there are questions you have to know the answer to that require new kinds of data or that you don’t really know how you’re going to do that. It reveals the extent of which you really understand how you’re going to treat your customer, how you’re going to price something, what your risk is whatever kind of decision you’re trying to make, and that enables you to do a better of it.
But it’s also a very focusing thing because a lot of people, for instance, will say,, “And I’ve got to make a decision at this point in my process. I’m going to assemble all the data that might be relevant and give it to someone. The problem with that is it’s a little bit of a blunt instrument, and if I model the decision, then to some extent, I can sort of finesse it. I can say, “Well okay, this is the data that actually seems to contribute to the decision-making the way we want to make the best practice, the way we identify, the way a good staff makes this decision.” That enables you to take a sort of ‘decisions first’ kind of mindset which says, this is the decision I’m trying to make, model, improve, and therefore this is the data I need. Rather than saying, here’s the data I’ve got, let’s hope it’s enough.
Joe: When I look at modeling, do I create the decision I want to out of it? How does modeling prevent me from doing that?
James: Well I can distinguish them; when you’re modeling these decisions, one of the critical things is you’re modeling a decision you’re going to make more than once. So this is when I go to process the model or a customer journey model. I’m not building it for a specific customer or a specific document process through my business. I’m defining a repeatable template that I’m going to apply over and over again. Decision modeling is just the same; I’m not building a model of how do I treat Joe. I’m going to say, how do I treat customers when they do ask. And when Joe comes, and he’s a customer. Then I can define how I’m going to decide what I’m going to do. Maybe I’m going to decide how loyal a customer Joe is, and I’m going to decide how valuable a customer Joe is, I’m going to decide what I think Joe is likely to do next and based on those things, I’m going to decide what offer I’m going to make Joe. But I can define that decision making approach in advance because I know I’m going to have to make it for lots of customers, not just Joe, and so it’s really about a template for my decision making approach rather than a model of a specific decision.
Joe: I’m not necessarily using it for an event or to solve a problem, I’m using it to develop a business process?
James: Yes, it’s very analogous to a process model. If I want to define the workflow through my business accurately, I go to a process model that tells me what the steps are, what the sequence is, who’s doing which piece, what the roles are. And then what we find is that within those processes, most processes have a number of decisions that have to get made, and what we see over and over again is that those are poorly defined in most process models. And so what decision modeling lets us do is say when we get to this point in the process, when we got three potential branches we’re going to go down, how are we going to pick which one we’re going to use for this customer, or this product, or this incident, or this quality problem? We can build a decision model that says, well here’s how we pick between these three paths, or here’s how we calculate the cost of this or other kinds of decisions.
Now we know exactly how that decision is going to get made and we may just give that to the decision makers and say, hey look guys, you guys are the analysts in the call center. This is how our best practice way of reviewing a claim for fraud or something. But you may also say, I’ve got a good enough handle on this decision model that I can write a bunch of rules for it and automate this decision. Or I can say, if that’s the decision these people are trying to make, maybe I can go to Dashboard that would help them make that decision. And so by being explicit about the decision-making, you create an opportunity to design new and more powerful computer systems to help make or automate that decision.
Joe: How difficult is it to get started? Do you need someone trained to use the process and do you need software?
James: So the first thing is, how difficult is it? Well, just like any modeling technique, the approach we use and the one, we recommend to people, is based on a new standard that has been published by the Object Management Group. The Object Management Group (OMG), is the people who own the Business Process Model and Notation standard (BPMN), and they’ve just published this new decision model and notation standard. So it’s a standard representation for decision models, in the same way, that BPMN is a standard representation for process models. Decision Management Solutions is one of the submitters backed by our big companies in the space, and it’s a very simple notation because it’s a very recursive approach. You break the problem down into pieces, so it has a very small path.
In a technical sense, it’s not that hard to learn how to read the diagrams or what the pieces are but like any modeling technique, practice, training and all of these things help you build better models. The difference between being able to build a model which you could probably do to be honest in about 10 minutes and the kind of working through an exercise, learning some of the tips and tracks and so on and when we do that as a training class, it’s like a day’s training in decision modeling and that’s enough to get most people up the learning curve and able to build reasonably good models reasonably quickly. But as always with a modeling technique, there’s no real substitute for experience and the more models you build, the better you get.
And about that software, now a lot of decision models are quite straightforward and early versions of the model can be done on a whiteboard or a flip chart and you can sort of have a discussion around it like you can with any modeling technique. But just like process models or data models, really at the end of the day, you’re better off with a repository based software tool that will let you capture that information so that it persists; so the next time you build a decision model and you want to reuse your definition of how do we decide how valuable a customer or somebody is, that definition is available to you. You don’t have to create it again from scratch. But yes, there is a key value proposition for software infrastructure that will help you manage these models as well as obviously for the kind of training and mentoring that any modeling technique tends to benefit from.
Joe: Is modeling at a point that you would put the software up on the screen, and it would prompt you to maybe ask the right questions or give you structure itself to build the model?
James: Most of them are not and certainly I mean in Decision Management Solutions, we have a software product for decision modeling. I would say if we think about sessions we run with customers, about half the time we would do the model on a whiteboard and then put it in the software later, and the other half of the time, we’ll use the software live. Most projects start with the first version; we sketch out on the whiteboard, we’re having a conversation, and then we’ll start to put the model in the software and then we’ll iterate the first few versions of the decision model, often very quite significantly as you get people thinking about the decision making. Then we’ll do the iterations in the software, and we’ll use it live in meetings and we’ll edit the diagrams live in meetings. The great thing about the notation is that it’s got very few shapes, very few lines. The pieces that you’re using to describe the model are individually quite straightforward and so we’ve had a great experience while we put a decision model up on our projector and we project in front a new audience who has never seen one before, and before we can even explain what the shapes mean, they’ve already discussing whether it’s right or not because if it’s your business problem, the model is transparent as to what it’s saying. It’s not obscure at all.
Given that, it’s actually quite easy to use the software live in a session. But no, they’re not quite at the prompting, suggesting stage yet. I think we’ll get there but the approach is not as well established as something like process modeling or data modeling, and so we’re not as far along in terms of patents and best practices as some of those other model notations are.
Joe: Well decision modeling to me kind of is like capturing my decisions. Explain the difference there to me a little bit.
James: If you think about a decision, if I’m trying to make a decision and I’m trying to make it this once, I want to decide if I should do business in a new country. Well, I don’t really need to model that because I’m just going to work through it. But if I’ve got a decision like what’s the next best action for this customer, if we have a banking customer, let’s say we’re doing a next best action initiative, they added up all their customers’ interactions, they have 140 million customer interactions a month. That’s a lot of customer interactions. You got to make that best action decision, 140 million times a month.
Well, you’re not going to do that randomly; at least you hope you wouldn’t, and you’re not going to do it manually. You need to be able to describe how you’re going to make that decision and then embed that into software so your software can make a good decision for you. That’s what you want a decision on; its the more repeatable, operational sort of transaction or decision. How do I price this order? Should I pay this client? How do I treat this customer? Those kinds of decisions are the ones that really pay in invest modeling because you’re going to make them over and over again. And so understanding how you make them is a critical first step.
Joe: What type of organizations use decision modeling? You mentioned banking there. What are some of the other users of it?
James: The biggest two industries right now are banking and insurance. They’re both heavily regulated; they’re both very decision-centered industries, and they make lots of decisions about trades, and accounts and claims, and policies, and everything else. So those are by far the most active. Most of our clients are in one of those two industries I would say. We’ve had other ones around manufacturing; we’ve got a couple in like Telcos and other places, and the challenge really is many of the people’s first thought about decision-making is about customer treatment. If you look into customer treatment decisions, you’re going to be focused in retail banking, in property and casualty insurance, in end user consumer telecommunications because those are the companies that have lots of end customers. But you’re beginning to see it in retail, in manufacturing, in consumer branded goods as they start to identify more of these decisions that they want to take more control of.
Take the retail for instance, there was a retailer I spoke to in Europe, and they more or less continually assess how many of each product they need to ship tonight to each of their stores. So if you’re making a decision about every one of these 10,000 products you stock, every 15 minutes, for every one of your 200 stores, you’re making a lot of decisions everyday about how many of these are we going to sell tomorrow, how many do we have. And so once you stop thinking about some artistic buyer making an informed choice and start thinking about it as a system. Then it’s a very high volume decision-making environment. So we’re beginning to see other industries recognize that they do make these decisions and they perhaps have hidden it by grouping everything together and saying, “Oh, we’re not a buyer for this range of products.” They make one decision, well they don’t make one decision, and they make lots of decisions about how many of this product go to this store. And as they recognize that, I think there’s a sense that they’re going to start modeling and managing more decisions. But so far, banking and insurance is clearly the number one and number two industries.
Joe: What is the level within the organization that’s used? It is always a big-picture map or can it be refined down to a much granular level?
James: No really, it’s almost the reverse. It really is at a granular level. It’s really about these high volume transactional decisions rather than about management or executive positions. A great Peter Drucker quote, he once said that the tendency to think that only executives who make decisions are the only executive decisions that matter; this is a dangerous mistake. And what we have found is that most organizations on their ability to hit their metrics, to hit their numbers, is highly dependent on decisions made really at the front line of their people, by their reps, by their store clerks, and by their systems. And decision modeling is really in those decisions. How do I better understand how we’re going to get those decisions made at the front line of the organization so that as the managers of the company, we have some control over that that we can’t exert any other way. We can’t see or have the visibility. We don’t have the ability to tap all those decision-making points in our company because there’s too many of them, but if we understood what the decision model was and we agreed that that’s how we wanted to make that decision, then we can embed that decision model or that decision approach into our information systems, well then we do have some control over how that decision gets made. It’s really those very granular, low-level decisions that get made based on decision models rather than high-level sort of strategic thinking stuff.
Joe: Would this help me and I’m going to take it from a Lean perspective here, I’m always under the opinion that if you understand and define standard work well, it makes it so much easier to make decisions because you know when you’re within your boundaries of making decisions and when you’re out of your boundaries. Does this help that though the process? Would this help me with it?
James: Yes, I think it would because it would let you describe how you think you make that decision. What are the pieces of that decision data are you using? What metrics helps you decide that it is within bounds or not within bounds? All those things can be very helpful, and it’s a more effective technique than trying to use a process model for instance. But the other interesting thing is what we found over and over again is an opportunity for what we call micro-decisions. A lot of organizations today, the way their systems work and their processes work, there’s a place in that system where that work thrown in that process where everybody or every transaction is treated the same. But in fact you could treat it slightly differently.
Take a checklist. Now I have a checklist for a certain step in my process. Well, we know that checklists work better when they’re very short and very specific. But, of course, being specific and being short are like contradictory. So what if I build a decision model that says how should I generate a checklist for this transaction and instead of having a standard checklist, I generate a checklist just for this transaction based on the facts I know about this transaction, the circumstances, who I’m talking to, who’s the person whose using the checklist; all those things could be factored into the checklist generation so I could make a decision as to what’s the right checklist for this person, for this task, at this moment. And those kinds of opportunities to really focus decision-making on a particular transaction, on a particular customer, on a particular task I think is one of the things that decision modeling really helps with because it lets you say, because of the I’ve got a way of describing the decision making I want, I’m not forced into doing it sort of mechanically the same for everybody. I can design in the flexibility I want but still know what our flexibility because I have a model of it. I’m not just giving up or giving people an Excel spreadsheet and hoping they’ll do something sensible. I’m still controlling it because I’ve got a model, but that model includes some flexibility in how we make decisions.
Joe: Your latest book I believe is the Micro Guide to Process and Decision Modeling. Is it really micro? I mean can you explain that title a little bit to me?
James: Tom DeBevoise, my co-author of the book, he has written a book called the Micro Guide to BPMN. It was a book about process modeling, BPMN, and it was called micro guide because it wasn’t intended to be like a huge reference to me. It was designed to be a fairly quick read, give you a good feel for BPMN rather than being a reference book, so it’s where micro guide came from. And then when the new decision modeling standard came out, he approached me and said, why don’t you take –well you know about decision modeling ND this new standard — and add some chapters to the book about decision modeling and then we’ll reissue the book. We thought it would be straightforward; I’d just write a few chapters, add them to the book, and we’d be good, right? But what we found is that once you start thinking about decisions and decision models, it often changes how you think about process modeling. You build a slightly different process model because you’ve got a decision modeling approach as well. What we found is that the book became much more of a combination of the two. Rather than a book about BMPM that referenced the new standard, it became a book that was genuinely about once you can model processes and decisions, how do you use these two techniques to model your business more effectively?
Joe: You also wrote a book, Decision Management Systems; where should I start if I want to learn more about the subject, with the micro guide or the other book?
James: I would say if you want to learn about decision management more generally, why we automate decisions, what are the technologies involved, use cases, examples of customers doing it, why this is a good idea, then Decision Management Systems, that book, The Practical Guide to Using Business Rules and Predictive Analytics is probably the best place to start. It’s a relatively easy read; it’s not a terribly technical book, and it’s got an overview of both the approach and the technologies and the sort of rationale. If you’re ready to start building decision models or you’re interested in decision modeling, then the new book is the right book. It’s a more focused book. Its only really about process and decision modeling, it’s not about the broader framework or about the technologies involved; it’s really just about how do I use process and decision modeling to describe a business problem.
Joe: Now you’re putting on webinars and some different ways that people can learn more about the process, correct?
James: Yes, that’s correct. We do a lot of webinars on decision management and on decision modeling, many of which I recorded and were available on our Website, which is decisionmanagementsolutions.com. It’s long but easy to remember. But also we have a training class coming up in January. We run online trainings in decision modeling and so we’ll be running a class in January on Introduction to Decision Modeling with the new standard. It’s like three one and a half hour sessions over three days. So we intended it to be easy to take, not disruptive to your day that you’d have to take the whole day out. It’s like you take a little time in three days, we record those, we make our software available to everyone who’s taking the class, and there are free trials so that people can build models as part of the exercises. So yes, online training, we do on premise training – that’s why I’m in India, and we also do a lot of webinars, a lot of white papers and written material about these topics. It’s a real passion of mine, and there’s a lot of that material up on the Website.
Joe: How often do you offer that training that you’re putting on January 6? Is that something that you do repeatedly or once a quarter or something?
James: We try to do it once a quarter. We’ve got a 1-day, 3-session on Decision Modeling, and a 6-session one on Decision Management and Decision Modeling, and we try to do one or the other every few months. To a large extent, it depends on my schedule. If I’ve got a lot of traveling scheduled, it’s hard to be somewhere I can do the training from but once I know what my schedule looks like, we can put them on the calendar. So yes, January is the next one, and I’ll suspect we’ll do one in February or March, sometime. We’ve certainly got a 2-day one on the calendar in April, so yes they come up pretty regularly.
Joe: Well I think you’ve contributed a lot to this space from what I’ve learned just in the short time. Is there anything you’d like to add that maybe I didn’t ask?
James: One of the things we’ve been excited about with decision modeling is the breadth of use cases for it. We started thinking this was a way as it is to design, do the requirements and design for automated decision-making systems. What we found on the journey was that there’re lots of things we do as companies that are really about decision-making. We build predictive analytics, data mining results, because we hope they will improve our decision-making. We build dashboards to help people make decisions. We send you reports and visualizations because we hope that the content of that report will help you make a better decisions. And over and over again, we’ve had experiences where if you can focus on the decision you’re trying to improve first, you can dramatically improve how you attack many different kinds of problems, whether those are writing a procedure manual, building a dashboard, building a report, building a brick analytic model, or trying to document the business rules you need to automate a decision.
Decision modeling is a remarkably robust technique that we’ve used very successfully for everything from very high level models just to sketch out how a group of people make a decision, right down to enormously detailed fine grain models that will allow us to precisely implement what we needed to do.
The thing I will encourage people to say is that decision making matters to your business, repeatable decisions are everywhere in your business and you don’t really – probably today you have very good tools for finding them describing them or sharing that understanding amongst the people who work in the organization, decision modeling is a powerful context for getting that understanding.
Joe: What’s the best way for someone to contact you?
James: There’s a lot of contact information on the Website. The easiest thing is decisionmanagementsolutions.com; that’s our Website. You can email at [email protected], or you can email me, I’m [email protected]. If you’re a blog reader, I have a blog jtonedm.com. But probably the company site is the best place to go. Email me or email the company and ask your questions. We have a free trial for the Software; we have lots of free material in terms of white papers and recorded webinars. Those are something that we’re excited about. We’re seeing a lot of real exposure and interest in it in the last few months as the standard gets published, and people start to realize it’s a real thing. Yes, we’re looking forward to it.
Joe: That sounds awesome James. I appreciate your time. Have a pleasant trip back to the States when you get ready. And this podcast will be available on the Business901 iTunes store and the Business901 Website. Thanks again James
James: Thanks very much Joe.
Lean Sales and Marketing: Learn about using CAP-Do