In the last 20 to 30 years, companies have faced significant changes in how they perform their day-to-day operations, and so have the analytics used to make decisions. According to James Taylor, CEO of the independent consulting firm Decision Management Solutions, the transformation is both technological and environmental -- from the rise and accessibility of better processing power to the self-service approach customers are beginning...
to accept and expect from businesses.
"The expectations of people are moving, and that’s creating a real sense that companies have to provide smarter systems so customers don’t have to be referred all of the time or wait for approval,” he said.
Taylor proposes adopting what he calls “decision management systems,” or a way of automating as many day-to-day decisions as possible using a combination of business rules management and advanced analytics, also the subject of his latest book Decision Management Systems: A Practical Guide to Using Business Rules and Predictive Analytics.
SearchBusinessAnalytics.com recently sat down with Taylor to discuss both the promise and the problems of his decision management systems approach.
What is a decision management system?
James Taylor: Fundamentally, it’s a system designed to automate and continually improve a high volume of repeatable business decisions. When you make decisions as a business, you obviously make strategic and management decisions, but every time you interact with a customer, a supplier or a transaction, you make lots of operational decisions. Those decisions are about a single customer, a single transaction, a single supplier, a single order. What [companies] have historically done is they’ve treated everybody exactly the same. … Decision management challenges that assumption to automate these decisions in a way that doesn’t just treat everyone the same.
For businesses interested in getting started with decision management systems, where should they begin?
Taylor: The best place to start is to think about the decisions they make on how they treat customers, because a company of any size has a lot of customers and they make a tremendous number of decisions about how to interact with customers. Most businesses today treat everyone the same or they rely on their front-line staff to make good decisions about how to interact with them. And that works to a point, but if you have thousands of people in call center, that’s hard to do guarantee. So I think it’s, what are the decisions you make that affect your customers, and that tells you which decisions are going to make the difference to your business objectives. They’re different for everybody.
What kinds of architecture or tools are needed to implement this kind of a system?
Taylor: First, it’s an approach. You have to be willing to build decision management systems and not just processing systems. But really you need a couple of things: You need to have moved to a service-oriented mindset at least. You don’t necessarily need to be done converting your legacy systems to SOA [service-oriented architecture], but you need to be thinking in terms of the services and components that do specific things for you. Secondly, you probably need to adopt a couple of technologies that you may or may not be using today. One is a business rules management system. Decision making involves a lot of business rules such as regulation, policy, best practices, know-how. You have to manage all that logic somehow, and code typically doesn’t work well with that. And then you need to be thinking about more kinds of advanced analytics -- data mining and predictive analytics -- so that you can turn historical data into analytic insight that can be consumed by a computer.
We’ve heard the advanced analytics skill set can be a scarce commodity. Are you saying businesses need to invest in the people who can do this kind of work?
Taylor: Yes, I think they do. And it is a scarce set. There are a couple of things I would say could mitigate that, though. With decision management systems, what you’re trying to do is build a system that applies these analytics. One of the comments often heard about pervasive analytics is, how do you make analytics pervasive. When you start talking about advanced analytics, they say, “if advanced analytics requires this skill set, surely that makes it harder to make advanced analytics pervasive.” But in fact the reverse is true. Advanced analytics gets embedded into systems. So I can make an advanced analytics model that predicts customer churn, for instance, pervasive in my call center without anyone in the call center having to know how to build a churn model. I don’t give them the tools to build churn models; I give them a system that uses churn models to help them make better choices when interacting with a customer.
There’s the other side: potentially replacing employee intuition when responding to certain situations. Is that a danger in automating these processes?
Taylor: Sure. It’s always a danger. The question is always, sort of, which employees are you talking about. If you think about a typical call center rep, someone in the call center getting calls from customers, do you want to remove the ability to hear in your voice that you’re angry? No, clearly you don’t. On the other hand, do you really want to rely on the call center rep’s ability to say, “Gosh, this customer’s profile looks like someone who has been profitable in the past and is going to remain profitable in the future as long as they keep this account open.” No, you don’t really want them to have to make that judgment. That’s something the system will do better.
There is a balance between the two for sure. I think often people are talking to businesspeople, decision makers, executives, about these kinds of systems and they’re like, “but you pay me to make these kinds of judgments.” Yes we do, but you’re not answering the phone when customers call; somebody else is. We want to balance that person’s judgment with your judgment. We want to embed some of your judgments in the system they’re using so that when they get a call from a customer, they’re applying the collective know-how, wisdom and data that you have as an organization. Clearly, you can go too far, and often you see a blend.
You’ll see a system that will say of the 75 offers you could make this customer, these three seem the most likely to be appealing and here’s why. And then allow the person on the phone to say, given the conversation I’m having and the tone in this person’s voice, I’m going to offer this one and not that one. They still get to participate in this conversation, but they don’t get an unbounded set of activity.
Dig deeper on Business intelligence best practices