Zehra Cataltepe is the CEO of TAZI.AI an adaptive, explainable Machine Learning platform. She has more than 100 papers and patents on ML.

Products Out Of Stock, Customers Silently Leaving

Yesterday afternoon, I couldn’t find eggs at the store where I usually shop for groceries. While there were food shortages during the Covid-19 pandemic, we still experience shortages of certain items here and there.

The recent Shoptalk and NRF conferences, along with industry reports such as Gartner’s Emerging Technologies and Trends Impact Radar: Artificial Intelligence in Retail, emphasize customer experience as the top priority for retailers. The report highlights the impact of AI in enhancing customer experience, improving supply chain management and reducing costs for retailers.

Product shortages make consumers unhappy. Even worse, some shortages cause customers to leave silently, and we may not even realize it. For one of our clients, we discovered that we could reduce out-of-stock by 4% by recommending what to deliver to small stores. Previously, the 4% out-of-stock caused unhappy customers who didn’t complain about shortages of their favorite snack and maybe switched to other stores or brands.

Manufacturers and distributors try to compensate for these problems by overproducing or overstocking, but in this economy, these are huge risks. I recently met with a manufacturer who ended up overstocking millions of dollars of refrigerators and has difficulty selling them now.

Existing Predictive Approaches To Demand Prediction

Manufacturers, distributors and retailers in the snack industry work closely together to ensure the smooth operation of the production and delivery chain. Accurately predicting supply, sales and demand is critical to meeting consumer needs and generating revenue at each stage of the supply chain, from production to retail.

Careful coordination and planning are required to maintain optimal inventory levels and meet consumer demand. Advanced technologies like automated warehouses, real-time inventory tracking and predictive analytics enable companies to forecast demand, optimize production and streamline delivery. As a result, the snack industry can achieve an efficient and responsive supply chain that benefits both consumers and businesses.

Approaches like time series prediction or machine learning can work well when there is a set business process, enough data about decisions being made, and patterns in the data. However, in today’s dynamic world, consumers’ needs and buying habits (e.g., online versus in store) keep changing continuously, making past data less representative of what may happen in the future. Although there might be petabytes of past data, for the current prediction problem at hand, there might be very little “relevant” data. You always need to keep an eye on data drift and quality. Companies need to be agile and change their ways of supplying the needed goods. This breaks the connection between predictive models and the actual business, and it usually takes months for data scientists to update the models to fit the new business needs.

Business Inclusive And Agile AI Predictions In A Dynamic World Of Consumer Needs

A business-inclusive and agile approach to AI can help predict consumer needs and supplies, and make more accurate predictions. By incorporating business expertise into prediction models, using pre- and post-processing to embed continuously changing business rules, and enabling the understanding of how models decide through explainable AI and interactive, dynamic and real-time dashboards, companies can ensure that their sales predictions are based on a thorough understanding of changing customer and market conditions.

As businesses advance in their AI journey, they need more accurate models and may end up creating different models for different portions, such as regions and vendors. In addition, the transition from predictive to prescriptive models usually requires other ML models that are used to make predictions about what-if scenarios. As a result, the number of AI models and the needed updates increase as businesses become more AI-enabled.

Easier-to-use machine learning platforms allow for the fast creation and updating of multiple models, learning continuously based on new data and also expectations of what the future might look like. By leveraging AI to make more accurate and responsive sales predictions, manufacturers and retailers can optimize their operations, improve their bottom line and ultimately provide better products and services to their customers.

What can you do now?

If you’re a manufacturer or supplier struggling to keep up with unpredictable demand and shortages, consider how you can take action. Instead of relying on overproduction or overstocking, which can be a huge financial risk in today’s economy, consider leveraging advanced technologies like predictive analytics and machine learning.

By analyzing your data and incorporating business expertise into your prediction models, you can gain a more accurate understanding of changing consumer needs and market conditions. This can help you optimize your operations, maintain optimal inventory levels and ultimately provide better products and services to your customers.

The key is to take a business-inclusive and agile approach to AI, one that allows you to learn continuously based on new data and changing expectations. By doing so, you can stay ahead of the competition and continue to deliver high-quality products and services that keep your customers happy and coming back for more. Manufacturers and suppliers should prioritize collecting the right data for business issues such as demand prediction and take action on existing data without waiting for huge infrastructure investments. Exploring the power of AI in demand prediction today can help make a difference in your business.

Zehra Cataltepe

Forbes Councils Member

Forbes Technology Council

Also Published on Forbes:  https://www.forbes.com/sites/forbestechcouncil/2023/05/18/using-data-and-ai-to-predict-demand-and-increase-customer-satisfaction/