In this post, we take a closer look at data models and why they’re important in decision management. This is a continuation of our series on what’s new in SMARTS™ Washington.
What are Data Models?
Data models are frameworks that define what data is important, their relationship to each other, and how they’re stored. Organizations typically rely on data models as their blueprint for designing a database. Data modeling is the process of creating a data model. Typically this process involves gathering business requirements from stakeholders, converting requirements into business rules, and translating rules into a data model.
Example Types of Data Models
While there are several types of data models, they usually fall under one of three categories: conceptual, logical, and physical.
- Conceptual models are the most high-level and involve identifying things (entities) that are important to the organization and how they relate to each other. For example, applicants and customers, where all customers are applicants, but not all applicants are customers.
- Logical models build upon conceptual models by identifying key details (properties) of each thing. For example, important details for an applicant may include their name, address, phone number, when they applied, and what they applied for.
- Physical models build upon logical models by providing instructions for how a database will store each thing and related details. For example, a database may store when an applicant applied as a timestamp with the field name “app_submit.”
Why is Data Modeling Important?
Data modeling and ultimately databases are important to organizations because they shape how organizations make, execute, and evaluate decisions.
- Creates Single Source of Truth: Data models provide consistency across departments and make the “executive decision” when ambiguity is at play. Even data such as name and address is not so straight forward as people change their names and move.
- Facilitates Institutional Memory and Knowledge Transfer: Data models preserve organizational knowledge and ensures continuity even when personnel changes occur.
- Enables Automation: Data models provide the consistency, structure, and context for AI and automated decision systems.
Best Practices for Decision-Centric Data Modeling
To ensure your data models are conducive to decision automation, here are some best practices when data modeling:
- Start with the decisions. Identify the key decisions that are being made across the organization and work backward to determine what data structures will be support them.
- Involve cross-functional stakeholders. Make sure there is good representation across departments from both business and technical domains.
- Document thoroughly. Maintain clear documentation so that decision-makers can understand the context and limitations of the data.
Data in Decision Management
In computer science, garbage in, garbage out is the concept that bad inputs (incomplete, flawed, biased) will yield bad outputs. The same can be said about automated decisions. Bad data will yield bad decisions. Therefore, it is crucial that organizations invest in data modeling and database management more broadly before investing in decision automation. Decision management platforms like SMARTS™ enable organizations to transform good data into automated decisions in a transparent, iterative manner. Unlike other platforms such as CRM or Marketing Automation which have their own data models (and requires you to map your data onto them), decision management platforms typically work with the organization’s data models. This ensures consistency from development to production and facilitates seamless data import and export across systems.
Data is Central to SMARTS™
SMARTS™ enables users to author decision logic in context of their data. Users can start a new project by simply uploading a file of test cases. If the data file explicitly includes the data model, SMARTS™ will import and save the data model separately (as a form). This form is searchable, version controlled, and shown in graphical form how it relates to other project assets. This enables users to easily keep track of changes and how those changes impact other assets.
In cases where the data model isn’t included with the data file, SMARTS™ will infer the data model by analyzing the data structure and types in the file and create a form. Alternatively, when no test cases exist, SMARTS™ enables users to create a form from scratch. To simplify data modeling, users can leverage AI Assistant, SMARTS™ generative AI tool, to iteratively create a data model through natural language. Users can then also user AI Assistant to generate test records based on the form.
Authoring Decisions in Context of Data
No matter how a form was created, users can edit or extend the form as they see fit. For example, users can define value constraints (value domains) for specific fields such as what product an applicant applied for (value must be either Product A, Product B, or Product C). The form and test cases will appear in the SMARTS™ user interface so that users can reference the data model while they author business rules and immediately test the outcomes on the test cases. SMARTS™ patented point-and-click rules authoring tool (RedPen™) allows users to simply click on the fields in the data model and follow the prompts to author rules in a transparent and intuitive manner. With SMARTS™, data models are the start and heart of every decision project.
Learn more about SMARTS™ Data-Powered Decision Manager.