Topic 3: Artificial Intelligence and Decision Support Systems

This section highlights the 5 stages of analytics maturity, the distinction between internal and external DSS data sources, the importance of balancing algorithmic systems and human retail intuition, and the 4-step decision-making process by Herbert Simon for inventory control.

The Analytics Maturity Ladder

By upgrading from paper and spreadsheet-based record keeping to an information system, a small retail company can climb the Analytics Maturity Ladder. Each tier provides strategic benefits, progressing from backward-focused reporting to automated recommendations as follows:

  • Descriptive Analytics (What Happened?):

    This involves the gathering of historical data about transactions using the Point-of-Sale (POS) system. For instance, creating an end-of-the-week sales report, which indicates that 150 units of organic coffee beans were sold.

  • Diagnostic Analytics (Why Did It Happen?):

    Examines the relationship of the data to discover the causes behind performance patterns. For example, matching sales records with weather and calendar information to determine that cold and rainy Fridays lead to a 35% increase in the sale of warm beverages.

  • Discovery Analytics (What Patterns Exist?):

    Makes use of pattern recognition to discover unexpected consumer behaviour without any previous assumptions about it. For instance, discovering that consumers who purchase specialized baking flour on Saturdays almost always buy premium butter with it.

  • Predictive Analytics (What Will Happen?):

    Uses forecasting models based on statistical analyses of past trends and seasonality to forecast future demand. For example, predicting that there will be a requirement for 300 units of coffee beans during the next long weekend rush.

  • Prescriptive Analytics (What Action Should We Take?):

    Uses algorithms and business logic to make automated optimal business decisions. For instance, automatically generating a purchase order to the supplier of coffee at a reduced price and setting the pricing of slow-moving products on shelves.

Decision Support Systems

A Decision Support Systems (DSS) refers to an interactive information system meant to assist management in decision-making concerning complicated but unstructured problems.

For its results to be accurate a retail DSS requires two types of data inputs; the first being Internal Data Input which includes real-time data arising from inside the business activities like point-of-sale transaction data, stock room data, past sales data, and cost of purchase by suppliers. And the second being External Data Input which includes data feed arising from the environment external to the business such as weather conditions, public school holidays, road closures in the area, and competitor pricing.

Balancing Algorithmic Systems with Human Retail Intuition

Whereas the DSS can crunch numbers much more quickly than a person can, it is meant to assist the retail manager rather than replace them. Recommendations made by the analysis of data should go together with human traits such as;

  • qualitative knowledge about suppliers an example being whether an old, local supplier has any current issues going on at their place such as strikes from their workers or factory work.
  • Ethics and community considerations like not price gouging when there are shortages in the area or severe weather situations.
  • And human customer interactions like personal interactions between the customers and the business.

Herbert Simon’s Decision-Making Process

For demonstration purposes, to showcase how a DSS operates in a retail environment, we will consider how to avoid stockouts during the holiday season:

  1. Step 1.

    The Intelligence Phase (Identifying Problem):

    This system scans the current stock level and identifies an anomaly, stock levels for profitable seasonal gift baskets are going down 40% faster than expected, hence, there is an impending threat of stockout.

  2. Step 2.

    The Design Phase (Identifying Alternatives):

    The DSS proposes three possible alternatives:

    • Choice A: Placing an emergency order from Supplier 1 (Expensive, but quick delivery within 24 hours).
    • Choice B: Placing an order from Supplier 2 (Least expensive, but a four-day delay).
    • Choice C: Transferring the available gift basket stock from a slow sister store.
  3. Step 3.

    The Choice Phase (Choosing among Alternatives):

    After reviewing the benefits and disadvantages of each alternative proposed by DSS, the manager would accept Choice C to avoid expensive transportation charges as well as to meet customer demands before the end of the week.

  4. Step 4.

    The Implementation Phase (Implementation):

    The store manager approves the transfer process in the system. This DSS modifies inventory levels in both stores and analyses the speed of sales in the next 48 hours to ensure the problem is solved.

Design and Pedagogy of Reasoning

Many small business owners often regard Artificial Intelligence and Business Analytics as complicated subjects meant for only big organizations. We developed this material in such a way that it demystifies AI and introduces the Analytics Maturity Ladder that explains how business owners can leverage from this technology right from the start. Our DSS explanation is based on Herbert Simon’s four-phase decision-making process, which shows the practical application of integrating internal and environmental factors in overcoming one of the biggest problems faced by small retail business owners – no need for inventory guesswork.