The Unspoken Cost of “Out-of-Stock”

Worldwide, consumer brands and retailers could gain $22 billion by recapturing sales lost when customers can’t find what they’re looking for, according to a report from the Grocery Manufacturers Association. Conversely put, the global cost of “out-of-stock” for brands is $22 billion — a figure that is predicted to double in the next few years. In the US, 15% of consumers who encountered “out-of-stock” switched to a different brand. Physical retailers lose $34.8 billion globally to their online competitors as a result of their customers missing the products they wanted. It has also been noted that 40% of the consumers who missed the products they wanted opted for substitute products from the same brand as opposed to resorting to a different retailer.

Most retailers look at Out-Of-Stock (OOS) as a metric that directly contributes to “lost sales” but as Raunak Gokhale puts it in one of his articles, “The complexity of the effects of OOS usually results into several other costs that are incremental to lost sales.” Some of them include inadequate ordering by distributors (resulting in low velocity SKUs being billed), sales employee time costs (grievance handling with retailer/distributor), underutilized costly space), consumer inconveniences and eventually loss of customers, decrease in forecasting/ordering accuracy, decreased store loyalty, increased likelihood for shopping at competitor’s store among many other things that eventually lead to the collapse of a business.

A research by ECR Europe & Roland Berger Strategy Consultants on global effect of stockout reveal results that every business should be weary of. When consumers encounter out-of-stock on the products they wanted:

• 37% of them will buy a different brand
• 21% of them will buy the same brand elsewhere
• 17% of them return purchased item later
• 16% of them will buy another item from the same brand,
• 9% of them will buy nothing

The same survey reveals worse effect when the cases of stock out occur multiple times:

• On the first occurrence of a stockout, 69% will seek a substitution while 31% will either not purchase or switch stores
• On the second occurrence of a stockout, 50% will seek a substitution with the other 50% opting for another store or no buy
• On the third occurrence of a stockout, only 30% will seek a substitution while 70% will either seek for another store or not buy.

We can see from these statistics that stockouts have a direct influence on customer churn and the more the cases of a stockout a customer experiences, the higher the probability of losing the customer to a competitor.

The more the cases of stockout a customer experiences, the higher the likelihood of losing the customer to a competitor.

When it comes to dealing with this issue of stockouts, the general rule of thumb has always been for stores to follow Pareto Principle: “Companies should aim to eliminate OOS for the 20% of items that account for 80% of total sales to make the greatest impact on the bottom line.” Whereas there are lots of factors that might lead to SKU stockout, one of the bigger ones being inefficiencies in supply chain management, the whole problem can be greatly improved by efficient and accurate demand planning. Companies not only face the challenge of stockout as a result of understocking but also face the challenge of overstocking leading to write-offs. The dilemma and the biggest challenge is thus for the retailer to get just the optimal supply of a product that will match the demand at any given time — the equilibrium quantity.

The equilibrium quantity of stock is the value at which the supplied quantity matches the customer demand. That is, all the stock purchased is sold on time before going bad and no customer encounters an out-of-stock on the product.

Over the decades, some companies have relied on sales forecasting as a method to help them anticipate the quantity of product sales so as to do their restocking based on the forecasted values. With simple tools like MS Excel, data analysts could build ARIMA models to help businesses achieve their sales forecasting needs but as we go into the next decades, the large sizes of data that companies have gathered over time means that tools like Excel are no longer capable of handling those large datasets; additionally, sales are affected by factors so many that classical models of the past decades like ARIMA have been passed by time due to their inability to model all the numerous factors affecting sales leading to largely inaccurate and hence unreliable forecasts.

Welcome to the era of cognitive demand planning where sales forecasts are data-driven and implemented through the power of machine learning. Cognitive Demand Planning at Insense Data Technologies, for instance, leverages the power of big data and machine learning to revolutionize demand planning and attain never-before-seen levels of accuracy. We not only monitor historical sales but also make use of all external factors affecting sales for every single product at every single store/location. Every product behaves differently and even the same product behaves differently from one location (store) to another in as far as consumer buying behavior is concerned. Weather patterns, for instance, is an important factor we consider in our modelling. Weather patterns have been known to affect sales of products but that knowledge only becomes useful if we use it in demand planning. For instance, extreme rains or sunshine would mean most people are in-doors leading to a general drop in sales but at the same time, would increase sales of certain products. Thus, the aim is to accurately predict sales for the next day knowing beforehand how the next day’s weather will be like.

The aim of cognitive demand planning is not only to identify the correlation of these factors with product sales but to use that information to attain accurate sales forecasts that would help improve the accuracy of demand planning — this is our mantra “from insights to actions”. The effect of holidays on sales is at the height of demand planning discussion and was actually the topic of our IDT Insights 2019 Report. At Insense Data Technologies, we have modelled all the holidays to ascertain how they affect the sales of every product and then use that information to better predict future sales. In Kenya, for instance, we have modelled a total of 28 holidays to help us understand how they affect sales of every product and with that, we can tell with extreme accuracy, how your sales will look like during the next Christmas festivity season. In addition to holidays, there are important events that affect sales eg the product having been on discount, or a promotion having been sent out or maybe it was a “back-to-school” week or an elections month or a Black Friday or even the general employment rate/average earnings of the people in that region among many other factors.

All these factors, intelligently put together, enables companies to achieve forecasting accuracies of over 95% on our Insense platform. Our aim is to help companies reduce the “out-of-stock” cases by over 50% and reduce stock wastage by over 70%. Now and more than ever, companies need to embrace cognitive demand planning and the end game, as we have already mentioned, is a myriad of benefits: increased demand planning efficiency, improved customer experience, improved customer retention, increased revenues resulting from the reduction of lost sales, reduced losses resulting from stock wastages as well as a general increase in returns on investment.