Decision Support System

Traditional DSS: Historically, DSS and BI tools relied on preconfigured, historical data with no ability to drive real-time decisions and action. With this approach, decisions are made based on the past.

Modern DSS: New tools and processes allow for “active intelligence”, a state of continuous intelligence with an end-to-end analytics data pipeline delivering real-time, up-to-date information designed to trigger immediate insights and actions.

Related resources:

• Learn 7 ways to accelerate data insights

Prior to decision support systems, organizational leaders relied heavily on a combination of their experience and professional training, and applied those to thoughtful use of the advanced insights generated by a data analytics platform. Decision support systems systematize that by taking organizational data, analyzing it, and presenting it for use in company decision making.

This DSS approach enables powerful augmented analytics or modeling to make analysis recommendations and game play the outcomes of different scenarios. By varying considerations, outcomes can be more accurately predicted and business decisions made based on the best available information. In this way, DSS supports both predictive and prescriptive analytics.

Three key elements that characterize a decision support system framework are model management, organizational data (your knowledge base) and user interface. Let’s briefly explore each.

Model Management: To make effective decisions, especially those made on an ongoing basis, it’s crucial for companies to develop key performance indicators (KPI’s) from which to evaluate performance against goals, and measure improvements over time. These KPI’s then form the decision criteria for the information models used to guide decision making. Having models provides the backbone of consistency every business needs to sustain success. Models can be leveraged by formally coded rules in DSS or prescriptive analytics software or by analysis using a BI platform.

Organizational Data or Knowledge Base: Before any DSS can be used, raw data must be transformed into clean, accurate, and up-to-date information. The graphic below illustrates how different types of data are combined, cleaned and transformed into standardized formats. The data is then stored in a repository such as a data lake or data warehouse using a governed data catalog.

User Interface: You’ve stared at enough dense tables of numbers to appreciate why it’s so necessary to have a more digestible and user-friendly way to consume data. A user interface, complete with digital dashboards, tables, graphs, widgets or other tools to present information, enables users to better interact with, view, and use the data at their disposal.

The many types of decision support system software options align to the various forms that data takes.

• Communication-Driven DSS: Helps companies manage data that requires more than one person to collaborate on a task.

• Model-Driven DSS: These DSS software solutions enable decision makers to access and manage statistical models used to run the business. The information is imported to a decision-making model that will then be used to analyze situations. These are the models mentioned above in the model management section.

• Knowledge-Driven DSS: Uses company data, facts, procedures, graphical displays, or pre-coded data rules to deliver informed solutions to specialized company scenarios.

• Document-Driven DSS: This type of DSS uses unstructured information available in various electronic formats in company systems, such as share drives, cloud storage, or other data asset management (DAM) solutions.

• Data-Driven DSS: Helps companies save, manage, and analyze a combination of data that’s both internal to the company and external to the company.

Now that we’ve defined what a DSS is, DSS characteristics, and types of decision support system solutions, let’s review decision support system examples to better contextualize DSS benefits across a range of situational applications.

DSS that Use Historical Data: Historical DSS data tabulates past performance and surfaces areas for improvement and/or provides a baseline metric from which to measure. These can include:

• Descriptive analytics: Sales, inventory, revenue-related figures.

• Diagnostic analytics: A combination of the “what” and the “why” behind data, providing the rationale from which to make go-forward decisions.

• Business Intelligence: A range of useful information used to guide important decisions (as described below).

BI Platform for Decision Making: Business Intelligence tools are a sub-segment of the larger decision support system definition, offering a range of insights, tools and data literacy benefits to organizations looking to expand data understanding, especially in the age of big data, AutoML, AI and machine learning. A major benefit of BI systems is that they are far less expensive and time-intensive to develop and implement than other DSS approaches.Augmented analytics refers to AI and machine learning processing and making recommendations on large volumes of data at lightning speed. Here’s a demonstration of how a modern BI platform allows conversational interaction and then suggests relevant insights.

Manual and Hybrid DSS: Before there were computer analyses, manual calculations, spreadsheets and assessments ruled the day. These drew upon available information and qualitative considerations. SWOT analyses are an example of this, as are analyses that rely on subjective judgment, such as performance reviews or creative/art direction. Hybrid DSS combines manual processes with DSS computational power of software applications to deliver a range of data for decision making.

Modeling DSS for Data-Driven Decision Making: This DSS example uses predetermined criteria to populate a query, then delivers the optimal solution based on the available data. This process includes the simultaneous evaluation of many different scenarios that are then presented for consideration. The better and more comprehensive the DSS model design, the richer the model outputs. To optimize models, two approaches are typically followed: rules-based and flexible optimization. The rules-based approach follows predetermined, predictable schedules, such as insurance risk tables. The optimization approach adapts to dynamic inputs and multiple constraints.

Using Predictive DSS for Future Decisions: These DSS tools tap predictive analytics techniques capable of anticipating future trends with a high degree of accuracy. By triangulating past events, current parameters and a range of external data too complex to do manually, predictive DSS can help companies make data-driven decisions about the future.