STOP building decision models that execute only a single type of decision; they must be composite and cover the full decision cycle.

April 18th, 2024

🛑STOP building decision models that execute only a single type of decision; they must be composite and cover the full decision cycle.

💥There are various types of business decisions, but almost all of them are complex and do not live in isolation. This nature of business decisions makes decision automation a very complex task.

Depending on the nature of the business decision, there are multiple choices to be used for decision automation:
1️⃣ Rule-driven decisions: These allow for the capture and execution of business rules around specific decisions using multiple methods, such as Decision Tables, Natural language, Tree-SubTree, and so on.
2️⃣ Data-driven decisions: This includes building a statistical model and data visualization for training and executing machine learning (ML) models to satisfy the specific decision's outcome.
3️⃣ Constraint-based decisions: Once a decision has multiple outcomes, there are techniques for modeling constraints to optimize a specific business KPI. Optimization models such as minimizing or maximizing something in the business can use constraint, linear, mixed integer, etc. models.
4️⃣ Decision-making process: These types of decisions look like processes, but the whole purpose of the process is to make a business decision rather than coordinate between people and systems.

⚠️Business decisions are complex. Therefore, a holistic view of business decisions requires embedding multiple of the above choices into a single hierarchical model. This is what it means to have a composite decision model. By decomposing a decision model into multiple smaller, easier-to-understand decision units, we can integrate multiple decision-making techniques into a single, hierarchical decision model.

⚠️Business decisions do not live in isolation. They require interaction with operational databases, reference data, customer data, historical data, ERP systems, processes, case management, and so on. These data are not mainly in the shape and form of information that the decision models need. Therefore, it is essential to access, shape, and form them to the structure that decision models require.

💣No #DecisionAutomation can scale if decision models are not composite to represent the holistic view of business decisions, and if they do not take care of data connectivity and real-time data processing and leave this messy procedural task to the consumers of the decisions.


First or last name is too short


First or last name is too short


First or last name is too short