Insights

Balancing accuracy and practicality in retail forecasting with AI

Peter Charness, VP Retail Strategy - UST

Managing optimal inventory levels is pivotal to retail success, and accurate forecasting is necessary to maintain the proper equilibrium between stocking enough products and minimizing surplus inventory. But how do you gauge the effectiveness of a forecast? Can it ever be too precise, and when does an emphasis on accuracy become counterproductive?

In the fast-paced world of retail, forecasting stands out as a critical business process that benefits immensely from AI. However, its value extends beyond mere number-crunching; it lies in generating actionable insights that drive tangible business outcomes.

Let's look at an example from a typical retail business.

DIVIDER

Navigating forecasting in retail

Imagine a scenario where a new analyst, Fred Forecaster, joins Great Grocery Company with a background in mathematics and a passion for optimizing forecasts, particularly in (his personal favorite) the cereal department. Fred dedicates considerable time and effort to refining his forecasting model using historical data and causal factors. His meticulous approach yields a consistently low 2% forecast error over 12 weeks—impressive by any standard.

Eager to demonstrate the potential savings from his improved forecasts, Fred presents his findings to company leadership, highlighting projected savings of over $7 million through reduced stock investments and improved in-stock positions. However, his excitement is met with a reality check during a warehouse tour.

While acknowledging Fred's precision, the warehouse manager points out practical limitations. For instance, despite Fred's forecast suggesting a need to ship most stores between seven and nine cereal boxes, every 4th day. However the warehouse can only ship in cartons of twelve due to packaging constraints. Handling partial cartons isn't feasible due to space and cost implications, negating the projected savings.

DIVIDER

Limitations in retail forecasting

Fred's experience illustrates common limitations in effective forecasting in retail. Traditional forecasting models, such as time-series analysis, depend on historical data to project future sales. However, past patterns do not reflect several contextual elements that impact current and future demand in a rapidly changing market environment. Relying solely on number-driven methods may lead to an oversight of real-world operational constraints in forecasting, just like in the case of Fred.

Pursuing precise forecasting must integrate a nuanced understanding of operational dynamics and market conditions. In making the forecast more intuitive, traditional models require help incorporating many external factors such as seasonality in demand, logistics, packaging, changes in affinity mix, and consumer trends. Given the scale and complexity of data, this can become an intricate process for analysts.

Advanced AI and analytics offer powerful tools to refine forecasts, yet their true value lies in their ability to inform decisions that are executable within existing business constraints.

DIVIDER

AI-powered retail: balancing accuracy and practicality in forecasting

Effective forecasting strikes a balance between precision and feasibility. It's not merely about generating accurate numbers but ensuring those forecasts translate into cost-effective and executable actions. With AI's powerful analytical and predictive abilities, retailers are equipped with an intuitive data-driven approach to assimilating practical, real-world elements in actionable forecasting.

Seen as a clear advantage over traditional methods, integrating AI in retail forecasting has several benefits. According to a McKinsey report, AI-driven forecasting can reduce errors by 20-50 percent, warehousing costs by 5 to 10 percent, and administration costs by 25 to 40 percent.

DIVIDER

A glance into some broad improvements in retail forecasting with AI

Actionable insights: Integrating AI in retail forecasting can be viewed as a shift from the traditional reactive approach to a more proactive one. AI tools can ingest vast amounts of data from diverse sources and predict supply chain disruptions. With ML algorithms trained to self-learn and continually improve predictions, retailers can refine their forecasting and reduce overstocking or stockout instances.

Real-time forecasting: Incorporating real-time data such as weather patterns, logistics tracking, etc., simplifies traditional methods that require constantly updating forecasting models.

Resource management: AI-driven forecasting enables agile responses to market changes that provide a significant upgrade from traditional methods depending on time-consuming and laborious manual adjustments. Automated algorithms can free up the manual bandwidth and shift it toward more strategic focus areas.

DIVIDER

The takeaway

In retail and across industries, the pursuit of precise forecasting must be tempered with a clear understanding of operational realities. While advanced AI models offer significant potential, their true value lies in their ability to drive decisions and actions that align with business constraints and objectives.

When refining your forecasting approach, remember Fred's lesson: the best forecast isn't always the most accurate one on paper but rather the one that can be effectively implemented to drive real-world efficiencies and savings. By embracing this perspective, retailers can leverage forecasting as a strategic tool to navigate uncertainties and optimize operations effectively.

Discover more about AI's transformative power in retail forecasting. UST's tailored AI and deep learning solutions provide the game-changing force to accelerate your growth and give you an edge over your competitors, positioning you as an industry innovation leader. Visit UST Alpha AI.

At UST, our teams redefine your retail experience, help you better understand your customers, and revolutionize your retail operations. Learn more about Retail Consulting Services and Transformation Solutions | UST.

RESOURCES

https://www.ust.com/en/industries/retail-and-cpg

https://www.ust.com/en/alpha-ai

https://www.ust.com/en/our-partners