The Need for Smart Enough Systems (Part 4): Operational Decisions
Last time, we introduced Enterprise Decision Management (EDM) as a systematic approach to automating and improving operational business decisions. In this instalment we expand on what operational decisions are and explain why and how so many of these decisions are hidden.
"A robust decision is the best possible choice, one found by eliminating all the uncertainty possible within available resources, and then choosing with known and acceptable levels of satisfaction and risk." 
Given that the EDM approach focuses on problems with operational decisions, considering some characteristics of these decisions can help you identify them more easily. If you look at your operations, the decisions made in those operations -- operational decisions -- are likely to have some or all of the following characteristics and be suitable for enterprise decision management:
- Volume. Perhaps the most common characteristic of operational decision problems is that the number of decisions of a particular type you must make is high -- so high you must automate it or high enough that many front-line workers must make it on your behalf. Volume alone can cause problems or exacerbate another decision problem, such as compliance or risk assessment.
What Is a Decision? When considering decisions, differentiating between types of decisions and the actual decisions is worthwhile. A type of decision might be "underwrite insurance policy" or "make cross-sell offer." An actual operational decision would be something like "underwrite James's insurance policy" or "make Neil a cross-sell offer." In this context, decisions have volume, not the types of decisions.
- Latency. Many managers now have more timely information about their business. If you can use this information to see trouble coming but can't change how you make decisions in time, you might have an operational decision problem. The latency between knowing something must change and being able to change it probably comes from having systems or people processes that are hard to change quickly. This is often caused by the way operational decisions are handled.
- Variability. Try imagining nightmare scenarios and thinking about what approach you might take. Think about the systems and people interacting with associates. Decisions those systems and people affect that must change to reflect your new approach could well be operational decisions that, although not a problem now, would cause problems if the business climate changed suddenly.
- Compliance. Ensuring that decisions are made consistently by using the same set of guidelines and policies and being able to prove to regulators that the correct rules are in place and used for a given decision can be difficult, especially if the decision must be made in any sort of volume. Demonstrating compliance in every operational decision can be particularly time-consuming if the decision isn't automated correctly.
- Straight-through processing. 'Straight-through processing' or 'once and done' processing involves performing every step in a transaction or process without human intervention. A manual review that drags down response time in a process might be hiding a problem-prone operational decision. If you have a mostly automated process that hangs up on manual reviews, you might have a good candidate for an operational decision.
- Managing risk. A prime reason for having a person involved in a process is to manage risk, which is often all about making decisions that manage trade-offs or risks and rewards. A risk-centered decision that must be made quickly or in volume might be a good candidate for an operational decision.
- Unattended. With some transactions, there's no choice but to automate a decision. Without automation, there's no way to inject expertise or learn what works better and improve the decision; for example, there's no person who can make a decision in transactions on your Web site or at your ATM. These kinds of decisions are often good candidates for operational decisions.
- Self-service. Complex decisions are more common in self-service. No longer is it enough for a self-service portal to deliver a document or ask someone to call an 800 number when things get complex. Now you need to automate this decision so that customers can self-serve, even when the decisions involved are complex.
- Personalized. Any time you want to personalize interactions with associates, you're making a decision. For most organizations, these operational decisions can create problems because of the need to balance timeliness with personalization.
Many decisions in your organization will have more than one of these characteristics. You might want to personalize a high-volume self-service interaction while taking fraud risk into account, for instance. One characteristic can be enough to consider using an operational decision, but combinations of factors obviously make a decision more problem-prone and likely to need a new approach to solve effectively.
One of the biggest challenges is finding the right decisions to automate and improve. Although most organizations can identify some operational, high-volume decisions that have the characteristics just listed, most make far more operational decisions than they think they do. These decisions are called an organization's hidden decisions. They are often well suited to automation, and finding them is a key step toward smart enough systems. Hidden decisions can be categorized as described in the following sections.
Next time we will explain some steps you can take to find hidden operational decisions. We will then go on to explore some techniques for finding operational decisions and show how applying enterprise decision management can deliver smart enough systems.
Probably the most common category is micro decisions. They are considered hidden because organizations recognize them as decisions but don't realize how many they actually make. For instance, if you decide to send a marketing letter to a subset of your customers, you might think you have made a couple of decisions -- what to put in the letter and who receives it -- and so you have. In addition, you have made a decision for each customer to receive or not receive the letter. Therefore, if you have 10,000 customers, you just made 10,000 micro decisions. If your Web site has 1,000 visitors each day, and you have decided to display a promotion, you have made 1,000 micro decisions to display that promotion to each visitor. When you add a new option to your interactive voice response (IVR) system, you have decided that everyone who hears that menu must have that option. When you decide on a product's price, you have decided to offer that product at that price to each potential customer who asks. You could consider the price separately for each customer, offering each a different price based on the value of the product to them and the long-term potential value of that customer to you. You should make micro decisions explicitly and thoughtfully, even if you decide to make them the same way for many associates.
Another way to spot micro decisions is to review your strategic decisions. If you have clearly articulated all the operational consequences in the strategic decision's description, you have probably described a number of operational or micro decisions that need to be managed. If you haven't described operational consequences carefully, you probably have hidden decisions. For instance, a strategic decision to improve customer retention needs a plan for making sure every interaction with customers that might influence their decision to stay or go has been reconsidered in that light. Trying to think through all the ways you hope the organization will change in response to a strategic initiative can reveal many hidden micro decisions that support the strategic decision.
Manual decisions are those hidden under manual processes. In other words, you might not think about decisions front-line workers make every day. Perhaps when your CSRs talk to customers about renewals, they are making an instinctive decision about retention risk before deciding what offer to make. Your delivery drivers might be making a decision about whether to accept a package for a certain delivery date based on their experience. Your store clerks might have to decide whether a customer can return an item without a receipt, or your salespeople have to estimate a prospect's price sensitivity.
You might want to recognize these decisions and leave them to your staff to make. Then again, you might not, or at least you want to think about how you could help them make better decisions that are more aligned with your corporate strategy, more visible to management, and perhaps more sophisticated and data-driven.
The third kind of hidden decision is what might be called a default decision. The policy controlling the decision was set a long time ago and never updated to reflect changing circumstances. Perhaps you established a return policy when all stores were independent, and you haven't changed it to reflect the more integrated approach used now that all stores are connected and supported by the same Internet infrastructure. Perhaps when your customer profile was quite different, you set up a policy for which claims are paid immediately and which ones are reviewed. Perhaps determinations of elite customer status or preferred supplier status are based on out-of-date views of these relationships.
These kinds of decisions often need to be revisited when technology and associates' expectations have changed. Default decisions can be out of date and be made at too high a level -- they could be hiding micro decisions, too.
Many decisions are made with only a short-term or narrow focus. For instance, when you decide to offer a customer a new credit product, you might be focused only on how likely the customer is to accept the offer and use the product. Your decision is driven only by the need to get the customer to use the new credit you're offering. If you consider a longer time frame, however, that decision might be based on the likelihood that a product is a gateway product that allows you to follow up with additional offers or the possibility that customers might not be able to pay their bills and will end up in collections. This longer-term decision often hides under a short-term one, especially because company objectives and employee performance might be measured only in the short term.
Long-term decisions aren't automatically better than short-term ones, but recognizing long-term decisions is important, and comparing costs and benefits of both short- and long-term decisions is often helpful.
Override decisions are hidden in a different way. When you look at the results of decisions made by using overrides, you don't get a true picture, because generally the logic isn't recorded. For example, your policy says turn down applicants age 18 or under. However, your staff sometimes approve applicants of that age who have wealthy parents. You might think this young customer base is a good segment to target, but you're looking at a truncated sample, and you don't know the 'rules' used to approve them. Similarly, when you draw conclusions about your customer base without considering that these customers have already passed a screening, your customer base isn't representative because it's composed of people who passed approval criteria or people who were already interested in your company's products. Clearly, manual decision making carries a risk that these override decisions will be hidden.
Missing decisions are those you don't think you can make. For this reason, you do the same thing every time. For instance, you have a one-size-fits-all approach to pricing because you don't think your associates would accept variable pricing. Perhaps you think your Web site and IVR system have to be the same for all users, or you include the same information in everyone's monthly statement. Failing to consider these decisions means that your associates do not get the personalized, focused service they deserve and that you do not get as many opportunities to influence them as you could.
One of the classic ways to hide a decision is to forget that not deciding is, in its own way, a decision. By not making a compelling or competitive offer or not doing so fast enough, you are, in effect, deciding to lose a customer or prospect. If your CSRs don't make a cross-sell offer, you're deciding not to do so. If you don't respond to information you learn about your customers, such as changes in their monthly deposit or payment patterns, you're deciding to do nothing. This category of decisions can be hard to spot and clearly overlaps somewhat with Missing Decisions. Although both are missed opportunities, making no decision implies an awareness of the possibility of making a decision and a failure to do so.
Although not really a hidden decision, considering conflicting decisions can often be worthwhile. If different parts of your organization treat associates in different ways, they are making conflicting decisions. Identifying, formalizing, and describing these differences can help you see what the real underlying decision is. Like the consideration of long-term decisions, a conflicting approach to a decision could be worse or better, but failing to consider and resolve the conflict 'hides' the decision.
Outside decisions are those made outside your organization. For instance, a customer decides to choose a certain channel for interacting with you, such as your Web site. If you could have influenced that decision, you might have made a different choice for that customer. Perhaps a competitor makes an offer to your customers. By accepting this offer, your customers cease to be a target for one of your products. In other words, other people, not you, are making the decision. Finding these kinds of decisions can help you determine whether you could do anything to influence them, and if so, what you could do.
These categories of hidden decisions often overlap and complement each other, but considering each category can help you find the hidden decisions in your business. Finding hidden operational decisions and clearly identifying those that aren't hidden becomes easier with practice.
|Acknowledgement: This material is from the book, Smart (Enough) Systems, by Neil Raden and James Taylor, published by Prentice Hall (June 2007). ISBN: 0132347962.|
# # #