Why should organizations model their important business decisions as part of digital transformation? We’ve been asked so many times to explain how our clients have benefited from decision modeling that we decided to capture it here. This article considers seven reasons to adopt decision modeling and summarizes the bottom-line benefits decision modeling has brought to companies that use it effectively.

The Importance of Decisions

A large part of conducting any business is making decisions. Some of these are strategic: should we enter a certain market, how should we design our new product, which partners and distribution channels should we choose? Others are more routine, made manually or automatically during everyday business operations. These are operational decisions.

Operational decisions include answering questions like:

• Should we extend a line of credit to this customer? On what terms?
• Should we initiate an inquiry into a customer’s insurance claim or just pay it?
• What products should we recommend to a client when they visit our website, given their past behavior?

Some operational decisions need to happen many times a second and so they are often automated. These decisions determine the day-to-day profitability of the business, how effectively it retains customers or how well it manages risk. Often the quality and consistency of decision-making determines your client reputation—for some clients their sole perception of your company is obtained from the outcome of these decisions. The logic of some decision-making is intellectual capital: it helps to establish or maintain a competitive advantage for your company—it represents what you do to better your rivals, your unique selling points. You need to make these decisions accurately, monitor their performance and manage their evolution. Digital transformation is partially about how this decision-making can be more customer focussed.

Part of a business analyst’s job is to identify, define and improve these decisions. That’s a challenge because often the details of the decision-making approach are trapped in the heads of subject matter experts, in reams of obscure documentation, in spreadsheets or within the code of software systems. These decisions may already be automated, but they are not leveraged because few people understand how they work, how effective they are or how to change them.

What is Decision Management?

Decision Management is a means of explicitly identifying and describing your business’s operational decisions—much as you would any other vital business asset (like data or process)—this allows you to ‘bring them out of the shadows’. It is an approach (and a technology stack) for describing, sharing, managing change and checking decision performance to see how they are contributing towards your enterprise goals. The goals of managing decisions are:

• To improve the transparency of decision-making by representing decisions in a format that allows business experts and IT to share a deep understanding of them.
• To increase the productivity of decisions by automating them in a disciplined way to increase throughput and reduce delays.
• To enhance the accuracy of decisions by supporting monitoring to measure their business effectiveness and collaboration to improve their definitions.
• To boost business agility and adaptability by ensuring that decisions are easy to change and that the full consequences of such change can be quickly understood. The fact that decision models are executable without need for translation into code further improves agility.
• To improve the consistency of business operations by ensuring that regardless of business process or channel, the same decisions are reused where appropriate.

An important part of this is modeling decisions.

What is Decision Modeling?

Decision Modeling (DM) is a part of decision management that focuses on representing business decisions using a standardized notation designed for use by business analysts and subject matter experts, rather than developers.

Through Decision Modeling, business analysts can build and share robust documentation of how their business decisions work. A decision model consists of two views: the Decision Requirements View and the Decision Logic View.

The Decision Requirements View describes how complex decisions are decomposed into simpler sub-decisions and exactly what data and business knowledge are required to support them. It shows the hierarchical structure of a decision.

An example is shown on the right. The decision (the yellow rectangle with a thick border) is broken down, showing its dependencies on sub-decisons (yellow rectangles), input data (green roundrects) and knowledge sources (blue rectangles with curvy bases).

Decision requirements diagrams are a powerful way of showing the integration of business rules, analytics, optimization models, business know-how and source data within an architecture. They facilitate all of these by providing a context for combining them into an integrated whole. Subject matter experts find that creating this view is an effective way of eliciting and consolidating their understanding of key decision-making.

The Decision Logic view (see example on the left) defines the logic of decisions. Often, but not always, it uses Decision Tables to represent how decisions determine their outcomes. This allows business analysts to describe exactly how decision-making is done in their organizations. Decision models can also represent logic in other formats (e.g., decision trees, text) or integrate with other means to make a decision (e.g., analytics, cognitive models and optimization engines).

Decision Models are not just about requirements. The precision of the decision logic view means that decision models can be directly executed. Therefore they can be tested against real business data and even deployed into productions systems without the need for developers to translation them into code. This model-driven approach to decision architecture reduces errors and allows for rapid change in decision-making—it provides safe agility for business decision-making and ensures that your specification of decision-making is always specific and consistent with the implementation.

Seven Reasons to Model Your Business Decisions

1 Managing Complexity, Maintaining Integrity

Unlike ad-hoc spreadsheets and even Business Rules, Decision Models are designed to scale effectively while remaining easy to understand. They achieve this by ‘divide and conquer’: separating complex problems into logical hierarchies of simpler concepts which can then be expressed and evolve independently, but yet collaborate explicitly and effectively.

A Decision Model is explicit about decision requirements: logical dependencies between business decisions and: other (sub) decisions, externally-defined policies, regulatory mandates, laws and business data. This makes it easier to determine the exact impact of changing decisions, data and externally defined policies. Many DM tools can illustrate the impact of making an internally- proposed or externally mandated change, for example a change in data standards or a regulatory compliance mandate.

Furthermore, DM gathers all related logic into one place: promoting logical integrity within rule sets, helping to identify gaps and inconsistencies within the logic as requirements are being formulated and thereby reducing errors from the start.

The explicit statement of dependencies engendered by decision modeling helps to identify real data requirements very quickly, avoiding over-provision of expensive data and the loss of productivity associated with late discovery of missing data.

2 Transparent and Precise Communication of Business Policy

One of decision modeling’s principle motives is to capture and communicate business decisions in a way that’s transparent to business subject matter experts. This allows them to understand, verify and even simulate decisions. DM provides a business-oriented representation that is separate from implementation details, yet precise enough to support execution. Explicit representation of business decisions liberates them from the bowels of IT systems and from the heads of subject matter experts, bringing them out into the light for all to see and avoiding loss of business expertise caused by legacy system decommissioning and staff attrition.

DM aims to increase the effectiveness of communication between business analysts and development teams. Decision definitions can be unambiguously shared and collaboratively reviewed so that fewer mistakes are made and any mistakes are rapidly identified and resolved. Time can be spent more productively in discussing the correctness and ramifications of requirements rather than resolving misunderstandings about their meaning. This time saving radically decreases time-to-market and error rates.

Decision models are also a very effective platform for training new staff in organizations’ decision-making logic.

3 So Precise, they are Executable

Decision models can define logic so precisely that they are executable. This is very significant because it means that these models are not just boxes, lines and text but the basis of a testable and deployable decision service. Executable decision models can become the prime record for decisions—making them a living specification of operational decision-making. They eliminate the need for IT to perform an error-prone translation of decision-making practices from dusty specification documents arcane spreadsheets and interviews with experts into a production system. Instead the decision model that business analysts and IT develop together can be simulated, tested and deployed directly.

Decision models can be associated with metadata that defines the true business value of a decision. For example, the business goals the decision is contributing to and how well it performed (e.g., how much money was saved, how much quicker was an outcome reached). This means when they are executed this performance can be continuously monitored and any undesirable trends quickly identified and addressed.

4 Agile Change Management

Explicit definitions of business decisions that can be understood and simulated allow business analysts to take direct control of the evolution of business policy, rather than having to intermediate through IT teams that own the systems that implement this policy. Analysts can correct mistakes and make improvements with little IT involvement (using simulation and impact analysis), deploying changes to a integration testing platform for testing by IT only when they are sure they are correct.

This thorough understanding of decision dependencies enables effective change impact assessments and leads to agile change cycles. These advantages cannot be provided by existing approaches like Business Rules.

5 Traceable Specification of Decisions

DM and business rules together support complete traceability from policy mandate to production system, allowing every outcome reached by a decision to be accompanied by a justification (in terms of the policy mandate) which can be examined in hindsight. This can assist in meeting regulatory requirements by demonstrating compliance with a regulation and help to diagnose faults quickly. This traceability can also streamline change management when regulatory mandates are updated because the parts of the mandates that change can be readily mapped to the parts of the decision model that support them and therefore need to be amended.

6 Standardization

DM standardizes the format of business operational decision requirements using the Decision Model and Notation standard (DMN). DMN is an open standard defined, ratified and published by a team of experienced vendors under the auspices of the Object Management Group Inc. (OMG). The OMG is an international, not-for-profit, technology standards consortium specializing in the development and custodianship of standard notations used by businesses in over twenty market verticals. Many of their standards are already well known to business analysts (e.g., BPMN, UML).

DMN provides a powerful and flexible notation and formal grammar for representing decision models and business logic. It has a simple, extensible and easy to learn structure which captures much of the essential information required to define decisions. It is designed for use by business analysts and business subject matter experts as well as IT professionals.

The use of such a standard means every modeler uses the same means to represent business decisions—so they can understand each other’s work even if they don’t have the same background. This is not only a big improvement over the ‘everyone has their own spreadsheet format’ free for all but also allows companies to hire new DMN-aware BAs in the knowledge that they will understand the models already created by others. This increases transparency: allowing effective review of business requirements by all stakeholders across the world, not just those in the inner team. It also promotes consistency between different decision authors, different teams and over time.

7 Simplification of Other Models

Frequently, aspects of decision-making are scattered across a range of text or spreadsheet documentation, hidden in business processes or hinted at by data models. By explicitly modeling decisions using a separate, standard model which integrates with business data, business process and business motivation models, decision modeling reduces the complexity of these other models, allows each to express those aspects of a system to which they are best suited and allows each to evolve at their own rates. DMN is designed to interoperate with BPMN and UML and is ratified by the same standards organisation.

Bottom Line

The bottom line is that Decision Modeling, by providing a transparent, traceable, scalable and precise representation of business decisions that is executable, that integrates analytic, optimization and cognitive technologies, that integrates decision making within a business process and that tests the dependencies that decisions have on business data and external authorities:

• Lowers time to market, by allowing business analysts to build, test and improve transparent models of decision making without specialist infrastructure or heavy IT involvement
• Lowers costs and improves reliability not only by supporting automation of decision making, but also by avoiding logical flaws and data provision errors
• Improves agility by allowing the full impacts of proposed changes to be seen before they are committed and by facilitating rapid amendments
• Manages complexity by supporting the hierarchical decomposition of processes and integrating many different logic representation styles into a coherent whole
• Improves business engagement by ensuring early and constant involvement by subject matter experts and business analysts in the evolution of business decisions
• Supports the continuous improvement of business operations by allowing the measurement of decision effectiveness and making them accountable for their consequences.

Practical Applications of Decision Modeling

Decision modeling is most frequently used in areas where high operational throughput, complexity or change are expected. Business areas where accountability and effectiveness is important also benefit. These include but are not limited to: risk assessment and management, pricing, regulatory compliance, KYC, eligibility, cross-selling, telecommunications tariffs and fraud detection.