≡ Menu

Managing a Decision Inventory

One of the most important assets for building Digital Decisioning is an up-to-date, accurate, and managed decision inventory. You need an understanding of the decisions in your business, especially those that are repeatable. A model of decisions and their dependencies, as well as their relationship with your performance management and business process environments, gives you a solid foundation for developing Digital Decisioning. Building out and managing such an inventory should be done business process by business process. Good decision inventories are built collaboratively by business, IT, and analytic groups. A decision inventory should be refactored as new information is gathered and kept synchronized with performance management changes. To maximize its value, it should also be linked to the implementation artifacts you develop as you build Digital Decisioning.

Business Process by Business Process

Some organizations like to create broad, across-the-board initiatives to standardize any good idea. They try to ensure a single enterprise information model, create an organization-wide process repository, or manage a single service inventory. There is nothing inherently wrong with this mindset—unless the objective is to create such an enterprise asset with a single project. It is extremely difficult to show the business value of an enterprise-wide standardization effort of this kind, and failure rates for these kinds of projects are high. While an enterprise-wide decision inventory is worth having, it is not a best practice to attempt to create the whole thing in one project.

The best way to build a decision inventory is to populate it over time. The decisions that matter to the organization can be identified business process by business process. The decision discovery necessary is conducted in the context of a specific project, focused on improving the decisions within a single business process or a tightly coupled set of processes. Once some experience has been gained in decision discovery, it is worthwhile to invest some effort in developing a framework to allow the decisions associated with each process to be integrated coherently and managed over time. By waiting until some initial experience has been gained, it is more likely that a manageable approach and suitable internal processes can be defined. By doing it reasonably early, it should not be difficult to retrofit existing decision definitions into the new decision inventory.

Create a high-level map
When first working on decision discovery, it is worth considering the different areas and business processes that will produce decisions to be captured. A wide but shallow view will help put each subsequent project into context and help keep the level of re-work to a minimum. Don’t allow this effort to get out of hand, though, as getting into too much detail across the whole enterprise will bog you down.


Decision discovery, as already noted, requires collaboration across business, IT, and analytic teams under the umbrella of a Center of Excellence, or CoE. Maintaining the decision inventory should likewise be a shared responsibility. It should primarily be a business responsibility in most organizations, though well-established analytic groups could perhaps take overall ownership. Regardless of who is “on point,” all three groups must be active participants. In addition, the various divisions and departments of the organization need to come together around the inventory. Decisions often involve multiple groups, with decision-making approaches defined by one organization, decisions made and delivered by a second, and all of these factors having impact on many. Keeping the various groups engaged in the decision inventory will improve its quality and make it easier to develop Digital Decisioning over time. Use coordinated review meetings and work with a CoE to make sure everyone stays engaged.

Refactor on each project

In an ideal world, each new project would reuse existing decision inventory content and add new content seamlessly. In the real world, the existing material won’t always be right for every subsequent project. The assumptions made when the initial decision discovery was done, the level of detail that was needed, or the understanding of the business behind the model could be completely erroneous. When the time comes to reuse that material in a new project, it will require some work before it can repurposed. When decision discovery is being planned for areas that overlap with the current decision inventory, build in some time to both re-use and “re-factor” the existing model of decisions. It is important not to dismiss the existing content and do it over. At the same time, it is unrealistic to believe that existing content can simply be picked up and immediately reused 100% of the time. Once the decision definitions have been reused a couple of times, however, they are likely to be stable and highly reusable in subsequent projects, as the friction of two or more different projects will take all the rough edges off.

Synchronize with performance management changes

One of the most important aspects of a decision inventory is its set of links to the organization’s performance management environment. The way in which specific decisions impact specific KPIs (key performance indicators) and metrics is important both for understanding good and bad decisions and for driving prioritization and design decisions in Digital Decisioning projects.

Because of this tight linkage, any change to objectives and measures should be considered to determine what impact it has on the decisions that support them. These performance management changes may be made to drive organizational behavior, to reflect a new acquisition or business strategy, or simply to better motivate employees. Whatever the reason for a change, it may well impact the way decisions are made—or, at least, it should. Once you have a model of the decisions that support a particular area of the business, it should be used to ensure that the way people and systems are making decisions is as aligned as possible with the way the organization measures performance.

For instance a company focused on growth may be emphasizing customer retention KPIs as critical to its success. Retaining customers, at more or less any cost, has become the norm. A change in focus to profitability means that only those customers making a contribution to the bottom line should be retained. Decisions about customer retention offers, among others, will now have to be changed.

Not all repeatable decisions are automated
While decisions in the decision inventory are likely to be overwhelmingly repeatable, they may not be 100% automated. It is not unusual, for instance, for most of the sub-decisions in a top-level decision to be automated, even though the top level is still a manual decision. Assessing the impact of a change in performance management approach should involve both looking at automated decision-making to see if it should change and informing those performing manual decisions.

Link to implementation

One aspect of a decision inventory is its ability to support impact analysis across the traceability of projects. To make this work, the inventory needs to have links not only to the business process and business performance models of the organization but also to decision implementation. If the inventory contains information about what implements each decision in the decision inventory, then it is possible to tell when decision-making approaches change, as any change to one of the implementation components represents such a change. Similarly, the need to change a decision means that the decision inventory can be used to identify all the implementation components that should be considered. Linking decisions to the business rules, predictive analytic models, and optimization models that implement them should be part of the decision inventory.