Retail stores lose billions of dollars in global sales because they usually do not have stock of products that customers want to buy in their stores. What are the consequences of this issue, and what can data science technologies offer to find viable solutions?

These are the possible causes of the shortage of mass consumption products on the shelves:

• Delays in the delivery of suppliers
• Absence of supplies for manufacturers
• Logistics problems
• Obstacles in imports
• Even the crisis generated by Covid-19
• Panic purchases also contribute to the shortage due to over-demand. 

Without going any further, during the first half of 2020, thousands of buyers worried about the effects of the lockdown and went to stores to buy flour, toilet paper, or gel alcohol. Despite not having too many obstacles in the supply chain, they returned home empty-handed.

A recent study by the IHL Group, carried out in 600 homes and businesses in the Retail sector, showed that buyers find out-of-stock products (known as OOS) in one each of every three shopping trips. Research by IRI showed that 20% of all out-of-stock remain unresolved for more than three days.

Thus, the average OOS rate is around 8% (one of thirteen products can’t be bought at the moment the customer wants to get it in the store). OOS hurts retailers and also companies of consumer goods, distributors, and product auditors who invest resources to detect stockouts.

Exploring Big Data solutions

Product stock problems have existed for decades, but the traditional solutions of companies to deal with them show various drawbacks, mainly due to the lack of accurate data analysis and implementation of reactive (and not proactive) solutions. How does Big Data have the possibility of power solutions with a predictive strategy?

Data science technologies allow production processes and the necessary stock to be adjusted to meet future demand.

It is about proactively using predictive tools to find the balance in the availability, without overproduction but without reaching shortages. A strategy that is essential for companies and large retailers to be more competitive since it allows them to plan, anticipate various market scenarios, and make decisions in the present that will bring benefits in the future. 

This process includes improving predictions of missing products, learning from historical situations of stock variations, and maximizing the value of consumption data, optimally redirecting logistics operations. However, it is always essential to monitor the data correctly and have an adequate strategy, which considers:

• Shopping Cart analysis: knowing which products are often purchased together can be complex to detect due to subtle changes in buyers’ behavior, but reveals trends. Predicting how each consumer approaches the buying cycle can help make strategic decisions about the assortment of certain items, their location, or how to optimize cross-selling.
• Analysis of product groups: it is about comparing the cause/effect relationships between sales and the benefits obtained, to plan storage strategies for those specific product groups.
• Demand planning: in addition to the historical record of sales, external data sources are considered for a satisfactory prediction, which includes some variables, such as the location of the stores, seasonality, socioeconomic characteristics, storage cost, storage times, delivery, and maximum stock, among others. Ultimately, this will help schedule and optimize supplier orders.
• Predictive and prescriptive analytics: the analysis uses a predictive analytics model, learning from past situations, and compares this with the current ones to predict what has already happened or to take action on what has already worked. The predictions will be valuable for stock management, allowing better planning. From these predictions, the best prescriptive analytics strategy is based on optimization, to simulate different scenarios and evaluate which actions and decisions will give the best results.

New challenges powered by collective intelligence and crowdsourcing solutions

What would happen if hundreds or thousands of consumers spontaneously reported the lack of stock of certain products in a collaborative app? Could companies, industries, or governments, in general, use this data for better planning?

Or, on the contrary, if these users reported on these platforms in which stores, in a specific city, a scarce or cheaper product can be found, in the event of an unfavorable situation, would they be helping other users with the same interest?

Citizen control combined with technology is still a viable option to project a near future, where the use of data is the result of the records left spontaneously and voluntarily by platform users.

The challenge for organizations is, without a doubt, to consider consumers as the principal source of information and to manage this valuable information, detecting hidden patterns that help to face the shortage of products, especially those of basic necessity.